Chemical Geology 234 (2006) 264 – 290

www.elsevier.com/locate/chemgeo

Magma chamber processes in the Tellnes ilmenite deposit

(Rogaland Anorthosite Province, SW Norway) and the
formation of Fe–Ti ores in massif-type anorthosites
B. Charlier ⁎, J.-C. Duchesne, J. Vander Auwera
Department of Geology, B20, University of Liège, B-4000 Sart Tilman, Belgium
Received 29 September 2005; received in revised form 5 May 2006; accepted 12 May 2006

Abstract

The origin of igneous Fe–Ti oxide ores associated with massif-type anorthosites is investigated through a detailed study of the
world-class Tellnes ilmenite deposit, part of the late-Proterozoic (930–920 Ma) AMC series of the Rogaland Anorthosite Province
(SW Norway). More than 100 samples from drill cores reveal significant petrographical and compositional variations within the ore
body. Four zones are defined, based on variations in modal proportions and cumulus mineral assemblages: the Lower and Upper
Central Zones and the Lower and Upper Marginal Zones. Plagioclase and whole-rock compositions discriminate the zones and
display patterns interpreted as a result of mixing of either plagioclase–ilmenite or plagioclase–ilmenite–orthopyroxene–olivine
cumulates with a melt of ferrodioritic (jotunitic) composition with a content decreasing from 80 to 20% from the margins to the
central part of the ore body. Phase diagrams for a jotunitic parental magma reproduce the crystallization sequence at 5 kb. The
orthopyroxene–olivine liquidus boundary is a peritectic in the Bjerkreim-Sokndal layered intrusion and a cotectic in Tellnes and
this explains the differences in the sequence of crystallization of the two intrusions. The high concentration of ilmenite, well above
cotectic proportions, resulted from gravity-sorting in the Tellnes ore body, which represents the lower part of a larger magma
chamber. Uniform Sr isotope ratios do not support magma mixing. The cryptic layering of the ore body precludes injection as a
crystal mush but favours in situ crystallization from an evolving magma in a sill-like magma chamber. The present trough-shape
and mineral orientations result from deformation during gravity-induced subsidence and by up-doming of the anorthosite.
Fractional crystallization of a TiO2-rich magma with ilmenite as an early liquidus mineral and plagioclase buoyancy are the
principal mechanisms responsible for the formation of Fe–Ti deposits in Proterozoic massif-type anorthosites.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Anorthosite; Fe–Ti deposits; Tellnes; Magma chamber processes; Cumulate; Trapped liquid

1. Introduction Tellnes ilmenite deposit (Rogaland Anorthosite Province,

Economic igneous Fe–Ti oxide occurrences are
commonly associated with massif-type anorthosites and
two hard-rock deposits are currently being mined: the
⁎ Corresponding author. Tel.: +32 4 3662250; fax: +32 4 3662921.
E-mail address: b.charlier@ulg.ac.be (B. Charlier).
0009-2541/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.chemgeo.2006.05.007
SW Norway) and the Lac Tio deposit (Allard Lake anor-
thosite, Quebec, Canada). The scarcity of Fe–Ti deposits
has hampered research on the formation of such ore bodies
and particularly on mechanisms responsible for ilmenite
enrichment. There is no consensus as to whether accu-
mulation processes, immiscibility, magma mixing, frac-
tional crystallization, solid-state remobilisation, cotectic
 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
crystallization, or a combination of these processes or
others, control the formation of Fe–Ti oxide ores.
Nevertheless, the genesis of massif-type anorthosites and
related rocks has been thoroughly debated (e.g. Ashwal,
1993; Longhi et al., 1999; Vander Auwera et al., 2000;
Bédard, 2001; Longhi, 2005) and the Rogaland Anortho-
site Province in particular has been the subject of numerous
investigations (see Duchesne, 2001 for a review). In this
province, the Bjerkreim-Sokndal layered intrusion, which
contains all rock types belonging to the AMC series
(Anorthosite–Mangerite–Charnockite), has been exten-
sively studied (see Wilson et al., 1996 for a review). The
petrogenesis of the Bjerkreim-Sokndal Intrusion has
greatly improved our understanding of the origin of
massif-type anorthosites and related rocks. An important
feature of the AMC series in Rogaland is the inferred
jotunitic (hypersthene monzodiorite) parental magma.
These liquids, also called ferrodiorites (Ashwal, 1993;
Emslie et al., 1994), are Fe-, Ti- and P-rich and, contrary to
basaltic magmas, crystallize ilmenite as an early liquidus
mineral. The liquid line of descent of jotunitic magma is
well known due to the occurrence of fine-grained jotunitic
dykes that represent liquid compositions (Duchesne et al.,
1989) and as a result of experimental studies (Vander
Auwera et al., 1998) performed on such compositions.
In this well-documented geological context, a com-
prehensive genetic model for the Tellnes ilmenite deposit
remains to be established. This deposit, which has been
mined since 1960 by TITANIA A/S, was previously
described as a rather homogeneous ore body with
plagioclase, ilmenite, and orthopyroxene as the major
minerals (Gierth and Krause, 1973). However, an
extensive database of whole-rock compositions built up
by TITANIA A/S over the last twenty years indicates that
compositional variations are larger than previously
believed. A detailed petrographical and geochemical
investigation of more than 100 samples from the deposit
has thus been undertaken in order to constrain a genetic
model for the ore body and Fe–Ti ore-forming processes
in massif-type anorthosites in general.
2. Geological setting
2.1. The Rogaland Anorthosite Province
The Tellnes ilmenite deposit belongs to the late-
Proterozoic Rogaland Anorthosite Province (see Duch-
esne, 2001 for a review), emplaced in the south-western
part of the Sveconorwegian orogenic belt (SW Norway)
(Fig. 1). The Province comprises different units of the
AMC series: four large massif-type anorthosite plutons
(Egersund-Ogna, Håland, Helleren and Åna-Sira), the
265
Bjerkreim-Sokndal layered intrusion (Wilson et al.,
1996), two smaller leuconoritic bodies (Hidra and
Garsaknatt) (Demaiffe, 1977; Demaiffe and Hertogen,
1981), minor intrusions (Vander Auwera et al., in press)
and dykes of the jotunitic (hypersthene monzodiorite)
kindred (Duchesne et al., 1989). Another characteristic is
the presence of numerous Fe–Ti deposits of subeconomic
to economic grade (e.g. Tellnes), occurring in the Åna-
Sira, Håland and Helleren anorthosites (Duchesne, 1999).
These magmatic rocks were emplaced between 930
and 920 Ma (Schärer et al., 1996) in granulite facies ortho-
and paragneisses, generating a wide, high-temperature
contact metamorphic aureole (Maijer, 1987; Möller et al.,
2002). Intrusion of the Rogaland Anorthosite Province
took place some 40 Ma after the Sveconorwegian orogeny
(Bingen and van Breemen, 1998; Bingen and Stein, 2003)
in a post-collisional regime (Duchesne et al., 1999). No
subsequent metamorphism has overprinted this magmatic
event. The anorthosite plutons crystallized over a large
range of pressures, starting at about 11 to 13 kb in large
magma chambers in the lower part of a thickened crust
(Longhi et al., 1993, 1999; Longhi, 2005). The low
density of a plagioclase crystal mush induced diapiric
uprise of the anorthositic massifs, possibly through zones
of weakness (Scoates and Chamberlain, 1997), to their
final intrusion at a pressure of around 5 kb (Ashwal, 1993;
Duchesne et al., 1999). Deformation observed in the
Province can be accounted for by anorthosite emplace-
ment (Barnichon et al., 1999) and gravity instabilities
(Bolle et al., 2002). Late doleritic dykes cross-cut the
Province and are related to the opening of the Iapetus
Ocean at 616 ± 3 Ma (Bingen et al., 1998).
2.2. The Tellnes ilmenite ore body
The ore body, an ilmenite-rich norite averaging slightly
more than 18% TiO2, was discovered in 1954 during an
aeromagnetic survey, and production started 6 years later
by TITANIA A/S as an open-pit mine. Reserves are
estimated to be 57 Mt TiO2, representing 14% of the world
reserves of ilmenite and 12% of the total (ilmenite + rutile)
world reserves of titanium minerals. By-products are
magnetite and Ni–Cu sulphide concentrates.
The Tellnes ore body is intruded into the central part of
the Åna-Sira anorthosite (Gierth and Krause, 1973; Krause
et al., 1985) (Fig. 1). This anorthosite body is essentially
made up of granulated plagioclase with rare isolated Al-rich
orthopyroxene megacrysts. Some parts are leuconoritic,
containing plagioclase, orthopyroxene and ilmenite. The
massif contains numerous Fe–Ti deposits which have been
mined in the past. The most important of these occurrences
are Storgangen, Bøstølen, and Blåfjell (Fig. 1) (Krause and
266 B. Charlier et al. / Chemical Geology 234 (2006) 264–290

Fig. 1. Geological map of the Rogaland Anorthosite Province (after Bolle, 1996) showing the location of the Tellnes ilmenite deposit in the central
part of the Åna-Sira anorthosite. Bs: Bøstølen intrusion; Bf: Blåfjell deposit; S: Storgangen; T: Tellnes “main dyke”; T’: Tellnes ilmenite deposit. Late
doleritic dykes are not shown.

Zeino-Mahmalat, 1970; Krause and Pape, 1977; Krause et
al., 1985; Duchesne, 1999). Mining is now confined to
Tellnes (Fig. 2). This deposit crosscuts the anorthosite and
thus seems to be younger than its host; U–Pb ages of zircon
is 920 ± 3 Ma for the ore body and 932 ± 3 Ma for the
anorthosite (Schärer et al., 1996). However, Charlier et al.
(submitted for publication) have concluded that the 920 Ma
age for Tellnes reflects the timing of Zr exsolution from
ilmenite and not the age of crystallization, which is
probably closer to the age of the anorthosite of around
930 Ma.
The deposit has a sickle-shape oriented WNW–ESE to
NNW–SSE and has a maximum width and length of
400 m and 2700 m respectively. The intrusive character of
the ore body is demonstrated by sharp contacts with the
host anorthosite, numerous apophyses and anorthosite
xenoliths, particularly abundant in the south-eastern part.
The main body extends to the NW and SE as a dyke (the
“main dyke”) with a jotunitic to charnockitic composition
(Wilmart et al., 1989). Dated at 931 ± 5 Ma (U–Pb ages on
zircon; Schärer et al., 1996), the dyke is not comagmatic
with the ore deposit, as shown by different Sr isotope
ratios (Wilmart et al., 1989). The two intrusions were
successively emplaced in the same structure, produced by
a transcurrent, dextral opening of a WNW–ESE-striking
zone of weakness in the Åna-Sira anorthosite (Diot et al.,
2003). Two sub-vertical doleritic dykes oriented WNW–
ESE cut the Tellnes deposit.
3. Previous studies
Detailed petrographical descriptions of the Tellnes
ilmenite deposit have been presented by Gierth (1970),
Gierth and Krause (1973, 1974), Gierth et al. (1982) and
Gierth (1983). Although these authors describe the
mineralogical and chemical composition of the ore body
as very uniform, they noticed some variability. Most
important are the lower ilmenite and the higher
 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
Fig. 2. Geological map of the Tellnes ilmenite deposit and the host Åna-Sira anorthosite at present level of exposure (after Krause et al., 1985). Grid is
the EURF89 kilometric UTM. Minor late faults are not shown.
plagioclase and apatite contents at the margins of the ore
body. The upper part contains increasing amounts of
magnetite and sulphides whereas the lower part is almost
magnetite-free and poor in sulphides.
Wilmart et al. (1989) reported a detailed geochem-
ical study (Sr isotopes, major and trace elements) of the
Tellnes dyke and the ore body. They inferred that the
ilmenite-rich norite could have been derived by
crystallization of a noritic liquid related to the host
Åna-Sira anorthosite. An emplacement mechanism
involving injection of a noritic crystal-mush, lubricated
by 3 to 10% trapped liquid, was proposed by Wilmart
et al. (1989). In this model, the ilmenite enrichment
was explained by dynamic sorting of dense ilmenite
during movement of the mush and its progressive
accumulation to form the ilmenite-rich norite in the
more central part of the conduit. The less dense
267
plagioclase-rich part of the mush was emplaced at the
margins and in the SE apophyses of the ore body.
Emplacement as a crystal-mush was later supported by
magnetic and petrofabric evidence (Diot et al., 2003).
The shape orientation of minerals was interpreted to
have resulted from flow during injection of the ilmenite
norite mush. From the well-constrained plunge of the
lineation (averaging SE 18°), Diot et al. (2003)
concluded that the mush was derived from an adjacent
(unexposed) magma chamber and was injected laterally
and slightly upwards from the SE into a zone of
weakness.
Robinson et al. (2003) recently compared the Tellnes
deposit to ilmenite-rich cumulates of the Bjerkreim-
Sokndal layered intrusion. In the latter, magma mixing
is inferred and the hybrid magma appears to have
crystallized relatively large amounts of ilmenite (Wilson
268 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
et al., 1996). Comparable compositional ranges of
plagioclase and orthopyroxene in Tellnes and in
Bjerkreim-Sokndal have led Robinson et al. (2003) to
suggest a similar origin for the two types of cumulates:
comparable parental magmas and possibly similar
mixing events but within chambers of different shapes.
Nevertheless, these authors admit that further documen-
tation based on more detailed mineralogical and
petrological studies is needed.
Extensive drilling by the mining company and
microgravimetric analyses have resulted in a detailed
picture of the 3D-shape of the ore body (Fig. 3). A huge
database of whole-rock compositions has also been
compiled by TITANIA A/S (Kullerud, 2003). About
5000 samples have been analysed for TiO2, P2O5, Cr, S
and modal magnetite. Of these, 2000 samples have been
analysed for all major elements and selected trace
elements. Based on compositional variations within the
ore body, Kullerud (2003) recognized four zones: a Lower
Central Zone (LCZ: ilmenite-rich, plagioclase- and Fe–
Mg silicate-poor), an Upper Central Zone (UCZ: ilmenite-
and Fe–Mg silicate-rich, plagioclase-poor), a Lower
Marginal Zone (LMZ: plagioclase-rich, ilmenite- and
Fe–Mg silicate-poor), and an Upper Marginal Zone
(UMZ: plagioclase- and Fe–Mg silicate-rich, ilmenite-
poor). These observations have, however, not resulted in a
new interpretation of the origin of the ore body.
Fig. 3. 3D view of the Tellnes ilmenite deposit constructed from drill cores and microgravimetric data with level of exposure before mining activity
(modified after Diot et al., 2003). Doleritic dykes not shown in 3D.
4. Samples and analytical methods
More than 100 samples were selected from various
drill cores in three SSW–NNE sections (800, 1200 and
1600), more or less perpendicular to the elongation of
the ore body (Fig. 4). Each sample corresponds to
40 cm to 1 m of a split drill core. Altered portions of
rocks have been avoided during sampling. A few
rocks were also collected in the open-pit. Petrographic
studies were carried out on polished thin sections for
all rocks.
Forty seven samples were selected from sections 800,
1200 and 1600 for plagioclase separation (Fig. 4).
Mineralogical separation of plagioclase grains (60–
150 μm) was performed using flotation in bromoform
and a Frantz isodynamic magnetic separator. Final
purification was carried out by HCl leaching to dissolve
any apatite grains and by hand-picking under the
microscope. Samples were ground in an agate mortar
and analysed for major elements by XRF on Li-borate
fused glass and for Sr and Ba by XRF on pressed powder
pellets (Table 1).
Sr isotopic ratios were measured for twenty one of
these plagioclase samples on a Fisons VG Sector 54 mass
spectrometer using dynamic multicollection at the Uni-
versité Libre de Bruxelles (Table 2). Prior to separation
using ion exchange resin, plagioclase separates were
 B. Charlier et al. / Chemical Geology 234 (2006) 264–290 269

Fig. 4. (A) Location of the three selected cross sections on the geological map of the Tellnes ilmenite deposit (same legend as Fig. 2); (B–D) Spatial
distribution of selected drill holes and samples in cross sections 1600, 1200 and 800. Bold numbers in italics are samples from which plagioclase has
been separated. The dotted bottom of the intrusion in section 800 has not been intersected by drill cores but is interpreted.

dissolved in a mixture of HF, HNO3 and HClO4. During 5. Petrography

the period of measurements, 87Sr/86Sr for the NBS897 Sr
standard was 0.710273 ± 0.000007 (normalized to 86Sr/
88
Sr = 0.1194).
Whole-rock compositions for 97 samples were
analysed for major elements by XRF on Li-borate fused
glass (Table 3). All the XRF analyses were performed on
the ARL 9400 XP at the University of Liege (Belgium).
Accuracy and reproducibility were evaluated on the basis
of a collection of international standards (Bologne and
Duchesne, 1991). Modal proportions have been deter-
mined by point counting (2000 points per 3 × 4.5 cm thin
section). Modal proportions are expressed as wt.%,
converted from the volume% using appropriate mineral
densities (Table 3).
5.1. Mineralogy and textural relationships
The Tellnes ore is an equigranular, massive, medium-
grained (∼ 0.5–2 mm) ilmenite-rich norite. The norite
rarely displays modal layering and appears extremely
homogeneous in outcrop. Plagioclase, ilmenite and
orthopyroxene are the main minerals. Olivine, magne-
tite, clinopyroxene, biotite, apatite and hornblende are
subordinate. Traces of Fe–Ni–Co–Cu sulphides are
ubiquitous. Other accessory phases include aluminous
spinel, baddeleyite and zircon.
Plagioclase is the most abundant mineral in the ore. It
forms elongated laths ranging from 0.5 to 2 mm in
270 B. Charlier et al. / Chemical Geology 234 (2006) 264–290

Table 1
Composition of plagioclase separates (XRF analyses) from the Tellnes ilmenite deposit
Sample Depth Zone SiO2 TiO2 Al2O3 Fe2O3 MgO CaO Na2O K2O P2O5 Total An a Or b Sr Ba
P8V-4N 238 UCZ 57.26 0.10 26.96 0.19 0.14 9.51 5.73 0.54 0.01 100.45 47.8 3.1 1170 169
P8V-6N 67 UMZ 58.53 0.05 26.15 0.21 0.13 8.62 6.11 0.72 0.01 100.54 43.8 4.2 1103 231
P8V-2S 16 UCZ 58.27 0.08 26.19 0.18 0.08 8.65 6.29 0.53 0.01 100.29 43.2 3.1 1162 212
P8V-2S 65 UCZ 58.52 0.07 26.35 0.20 0.08 8.72 5.88 1.01 0.01 100.85 45.0 5.8 1051 284
P8V-2S 158 UCZ 57.75 0.09 26.64 0.21 0.25 9.12 5.96 0.54 0.01 100.58 45.8 3.1 1147 168
P8V-7S 140 LMZ 57.98 0.09 26.75 0.21 0.08 9.22 5.91 0.58 0.01 100.84 46.3 3.4 1143 195
P8V-7S 180 LCZ 57.81 0.04 26.67 0.20 0.22 9.20 5.96 0.53 0.01 100.65 46.0 3.1 1156 180
P8V-8S 120 LMZ 58.70 0.04 26.39 0.20 0.09 8.73 6.14 0.62 0.01 100.93 44.0 3.6 1153 204
P8V-9S 44 UCZ 57.96 0.06 26.43 0.28 0.35 8.96 5.92 0.74 0.01 100.72 45.5 4.3 1098 225
P8V-9S 130 UCZ 57.32 0.06 27.04 0.20 0.11 9.55 5.74 0.54 0.01 100.58 47.9 3.1 1170 186
P8V-9S 187 LCZ 57.32 0.04 27.08 0.23 0.21 9.64 5.59 0.64 0.01 100.77 48.8 3.7 1155 206
P8V-9S 240 LCZ 56.96 0.05 26.91 0.28 0.56 9.54 5.54 0.58 0.01 100.44 48.8 3.4 1163 187
P8V-9S 298 LCZ 57.50 0.04 26.88 0.22 0.38 9.48 5.67 0.55 0.01 100.74 48.0 3.2 1152 180
P8.2V-9S 69 UCZ 57.15 0.06 26.77 0.25 0.24 9.39 5.73 0.55 0.01 100.16 47.5 3.2 1151 179
P12V-2N 104 UMZ 58.40 0.04 26.03 0.26 0.16 8.60 6.03 0.86 0.01 100.40 44.1 5.0 1068 252
P12V-13N 10 UCZ 56.41 0.06 27.00 0.28 0.15 9.34 6.03 0.59 0.02 99.88 46.1 3.4 1153 181
P12V-13N 140 LMZ 58.06 0.04 26.49 0.21 0.07 9.02 5.89 0.73 0.01 100.53 45.8 4.2 1118 206
P12V-3S 29 UCZ 56.40 0.06 26.75 0.20 0.14 9.23 6.06 0.63 0.02 99.49 45.7 3.6 1142 187
P12V-3S 87 UCZ 55.88 0.06 27.09 0.23 0.11 9.55 5.88 0.53 0.02 99.35 47.3 3.0 1144 158
P12V-3S 154 UCZ 56.28 0.06 26.98 0.21 0.11 9.45 6.00 0.54 0.02 99.65 46.5 3.1 1153 162
P12V-3S 230 LCZ 56.08 0.05 27.14 0.19 0.08 9.65 5.98 0.51 0.02 99.70 47.1 2.9 1182 161
P12V-3S 285 LCZ 56.00 0.04 27.01 0.24 0.06 9.63 5.83 0.64 0.02 99.47 47.7 3.6 1162 199
P12V-9S 77 UCZ 57.20 0.06 26.91 0.28 0.20 9.48 5.68 0.57 0.01 100.40 48.0 3.3 1166 191
P12V-10S 70 UCZ 56.07 0.06 26.83 0.22 0.10 9.29 6.02 0.57 0.02 99.18 46.0 3.3 1166 184
P12V-10S 110 LCZ 57.35 0.05 26.88 0.26 0.29 9.34 5.73 0.61 0.01 100.53 47.4 3.6 1150 185
P12V-14S 47 LMZ 58.92 0.03 25.84 0.24 0.21 8.25 6.12 0.88 0.01 100.51 42.7 5.1 1048 271
P12V-14S 75 LCZ 56.91 0.05 26.70 0.21 0.14 9.04 6.34 0.50 0.02 99.91 44.1 2.8 1133 154
P12V-14S 96 LMZ 57.22 0.04 26.29 0.27 0.20 8.63 6.27 0.81 0.02 99.75 43.2 4.6 1092 244
P12V-7L 10 UCZ 57.40 0.06 26.90 0.31 0.27 9.38 5.78 0.53 0.02 100.66 47.3 3.1 1148 170
P12V-7L 47 UCZ 58.23 0.05 26.44 0.26 0.15 8.92 6.09 0.63 0.01 100.79 44.7 3.6 1123 192
P12V-8L 40 UCZ 57.80 0.06 26.80 0.21 0.11 9.26 5.93 0.52 0.01 100.71 46.3 3.0 1165 178
P12V-8L 80 UMZ 56.59 0.04 26.87 0.22 0.09 9.28 6.03 0.70 0.02 99.84 46.0 4.0 1130 205
P12V-8L 82 UMZ 57.24 0.04 26.06 0.22 0.10 8.41 6.23 0.99 0.02 99.31 42.7 5.6 1051 278
P12.5V-6S 180 LCZ 56.09 0.04 27.15 0.23 0.11 9.69 5.87 0.58 0.01 99.77 47.7 3.3 1162 181
P12.5V-6S 239 LCZ 56.45 0.04 26.69 0.22 0.07 9.26 6.17 0.56 0.02 99.48 45.3 3.2 1152 170
P12.5V-6S 264 LMZ 58.44 0.03 25.46 0.23 0.10 7.64 6.52 1.24 0.01 99.67 39.3 7.1 964 355
P12.5V-2L 5 UCZ 57.20 0.06 26.79 0.23 0.12 9.44 5.76 0.55 0.02 100.18 47.5 3.2 1146 182
P16V-3S 7 UCZ 56.36 0.06 26.88 0.22 0.16 9.43 6.01 0.56 0.02 99.70 46.4 3.2 1151 172
P16V-3S 47 UCZ 56.04 0.06 27.02 0.28 0.25 9.44 5.94 0.58 0.02 99.63 46.8 3.3 1166 181
P16V-3S 84 UCZ 55.93 0.06 27.10 0.25 0.17 9.51 5.90 0.56 0.02 99.50 47.1 3.2 1166 177
P16V-3S 127 UCZ 55.99 0.06 27.07 0.28 0.23 9.50 5.96 0.56 0.02 99.67 46.8 3.2 1168 173
P16V-3S 166 LCZ 56.12 0.05 27.00 0.21 0.27 9.52 5.92 0.55 0.02 99.66 47.1 3.1 1149 160
P16V-3S 208 LCZ 55.88 0.04 26.77 0.19 0.27 9.27 5.94 0.63 0.01 99.00 46.3 3.6 1141 187
P16V-3S 231 LMZ 56.27 0.04 26.83 0.21 0.26 9.33 5.93 0.64 0.02 99.53 46.5 3.7 1143 194
P16V-8L 73 LMZ 58.55 0.04 26.35 0.27 0.15 8.48 6.19 0.74 0.01 100.79 43.1 4.3 1111 227
P00-30C UMZ 58.82 0.04 25.99 0.26 0.15 8.44 5.96 1.10 0.01 100.78 43.9 6.4 1004 294
a
An = 100 Ca / (Ca + Na).
b
Or = 100 K / (Ca + Na + K).

length, commonly with a shape-preferred orientation.
Laths may be slightly deformed, displaying bent
twinning, undulatory extinction, granulation (primary
crystals are replaced by smaller recrystallized grains),
and lobate contacts with adjacent minerals, suggesting
subsolidus grain boundary adjustment. Similar defor-
mation microstructures have been detailed in the Poe
Mountain anorthosite (Wyoming) by Lafrance et al.
(1996). Plagioclase commonly contains numerous very
small needles of oxide minerals. Some grains also
include large equant grains of ilmenite. Plagioclase is
undoubtedly a cumulus mineral.
Ilmenite contains lenses of exsolved hematite. It is
usually interstitial to the other minerals which
 B. Charlier et al. / Chemical Geology 234 (2006) 264–290 271

Table 2
Sr isotopic compositions for plagioclase separates from the Tellnes ilmenite deposit
87
Sample Depth An% Or% Rb (ppm) Sr (ppm) Rb/86Sr 87
Sr/86Sr ±2σ (m) (87Sr/86Sr)920
P12V-13N 10 46.1 3.4 0.85 1153 0.0021 0.704503 ± 6 0.704475
P12V-3S 29 45.7 3.6 0.68 1142 0.0017 0.704542 ± 9 0.704519
P12V-3S 87 47.3 3.0 0.78 1144 0.0020 0.704502 ± 7 0.704476
P12V-3S 154 46.5 3.1 0.59 1153 0.0015 0.704492 ± 9 0.704473
P12V-3S 230 47.1 2.9 1.85 1182 0.0045 0.704465 ± 7 0.704405
P12V-3S 285 47.7 3.6 1.54 1162 0.0038 0.704452 ± 7 0.704402
P12V-10S 70 46.0 3.3 0.96 1166 0.0024 0.704478 ± 7 0.704447
P12V-14S 75 44.1 2.8 0.83 1133 0.0021 0.704497 ± 6 0.704469
P12V-14S 96 43.2 4.6 2.21 1092 0.0059 0.704458 ± 6 0.704381
P12V-8L 80 46.0 4.0 1.17 1130 0.0030 0.704439 ± 8 0.704400
P12V-8L 82 42.7 5.6 2.64 1051 0.0073 0.704507 ± 6 0.704411
P12.5V-6S 180 47.7 3.3 1.05 1162 0.0026 0.704427 ± 7 0.704393
P12.5V-6S 239 45.3 3.2 0.85 1152 0.0021 0.704396 ± 6 0.704368
P12.5V-6S 264 39.3 7.1 4.05 964 0.0122 0.704693 ± 7 0.704533
P16V-3S 7 46.4 3.2 0.92 1151 0.0023 0.704520 ± 8 0.704490
P16V-3S 47 46.8 3.3 1.17 1166 0.0029 0.704497 ± 6 0.704459
P16V-3S 84 47.1 3.2 0.80 1166 0.0020 0.704505 ± 6 0.704479
P16V-3S 127 46.8 3.2 0.89 1168 0.0022 0.704464 ± 8 0.704435
P16V-3S 166 47.1 3.1 0.61 1149 0.0015 0.704469 ± 7 0.704449
P16V-3S 208 46.3 3.6 1.18 1141 0.0030 0.704475 ± 6 0.704436
P16V-3S 231 46.5 3.7 1.20 1143 0.0030 0.704487 ± 7 0.704447
Initial isotopic compositions at 920 Ma.

nevertheless locally contain inclusions of ilmenite. The
interstitial habit of ilmenite results from the coalescence of
grains with variable orientation and displaying 120° triple
junctions. This texture is also due to subsolidus grain
boundary migration of primary ilmenite (Hunter, 1996;
Duchesne, 1996). Exsolved hematite does not display any
deformation, implying that deformation textures in the ore
body were acquired above the solvus temperature. As a
result of recrystallization, distinction between cumulus
and intercumulus ilmenite is difficult. However, the
presence of equant ilmenite grains within plagioclase and
orthopyroxene shows that ilmenite is a liquidus mineral.
Orthopyroxene is present throughout the ore body but is
more abundant in the upper part of the intrusion. Two
different textures are observed: (1) subhedral, prismatic
grains with well developed Schiller exsolutions (Fig. 5A);
(2) smaller anhedral grains that are interstitial to plagioclase
(Fig. 5B). In the lower part of the deposit, only interstitial
orthopyroxene is observed whereas both textures occur in
the upper part. This suggests that orthopyroxene is an
intercumulus phase at the base of the deposit and becomes a
cumulus phase up-section. Interstitial orthopyroxene
reflects crystallization from intercumulus melt.
A small amount of olivine (generally <5%) occurs in
the upper part of the ore body. It appears more or less
simultaneously with the prismatic orthopyroxene. Only a
few samples contain cumulus orthopyroxene without
olivine. Olivine is commonly interstitial to the other major
minerals but can locally be found included in orthopyr-
oxene. The small proportion of olivine favours an
intercumulus character. However, the existence of olivine
included in other minerals and its relatively primitive
composition (∼Fo80, Wilmart et al., 1989) force us to
consider olivine as a cumulus phase that appears
simultaneously with orthopyroxene. Close to the contact
with the anorthosite, olivine has commonly been replaced
by symplectitic intergrowths of orthopyroxene and
magnetite (Fig. 5C).
Magnetite, clinopyroxene, biotite, apatite and horn-
blende are considered as intercumulus minerals that
crystallized from trapped liquid. This interpretation is
based on obvious interstitial, space-filling relationships
relative to plagioclase, ilmenite, orthopyroxene and olivine
and on their relatively low modal abundances. Grains of
clinopyroxene at the margin of orthopyroxene probably
represent externally exsolved granules.
Some samples close to the margin with the host
anorthosite have unusual textures. The most spectacular is
a “dalmatian texture” in which large plagioclase and
orthopyroxene oikocrysts enclose numerous, smaller and
uniformly distributed ilmenite chadacrysts (Fig. 5E–F).
This texture clearly demonstrates the early crystallization of
ilmenite. Other marginal rocks (12.5V-6S, 264 and 12V-
10S, 64) have a fine-grained texture in which two pyroxenes
occur in roughly equal proportions, together with anti-
perthitic plagioclase, ilmenite and a high proportion of
apatite (Fig. 5D). These are interpreted as chills, similar to
those of jotunitic dykes (Duchesne et al., 1985, 1989).
 272
Table 3
Whole-rock compositions for major elements (XRF analyses) and mineral proportions (wt.%) of cumulates from the Tellnes ilmenite deposit
Drill core Depth Zone SiO2 TiO2 Al2O3 FeOtot MnO MgO CaO Na2O K2O P2O5 Total Mg# Plagio Ilm Opx Olivine Cpx Biotite
8V-2S 194 LCZ 32.20 17.25 13.57 21.39 0.14 4.69 5.09 3.78 0.57 0.31 98.99 0.180 45.9 41.7 7.5 0.0 0.4 4.4
8V-7S 180 LCZ 30.77 17.96 13.05 21.97 0.15 4.38 5.07 3.51 0.54 0.38 97.78 0.166 46.6 44.2 3.6 0.0 0.6 4.5
8V-9S 155 LCZ 29.39 19.60 12.95 22.72 0.13 5.04 4.74 3.68 0.44 0.21 98.90 0.182 47.3 45.8 4.4 0.0 0.0 2.5
8V-9S 187 LCZ 30.42 18.91 14.19 21.36 0.12 3.70 5.18 3.94 0.49 0.23 98.54 0.148 43.5 50.5 4.4 0.0 0.0 1.6
8V-9S 213 LCZ 28.92 20.10 13.68 22.50 0.13 3.65 4.97 3.89 0.46 0.21 98.51 0.140 47.8 46.1 3.5 0.0 0.0 2.7
8V-9S 240 LCZ 27.96 20.98 13.27 23.31 0.13 3.56 4.82 3.75 0.46 0.25 98.49 0.132 44.4 50.6 1.9 0.0 0.2 2.9
8V-9S 269 LCZ 31.24 18.15 14.41 21.01 0.12 3.58 5.36 4.17 0.52 0.26 98.82 0.146 45.9 46.0 4.3 0.0 0.5 3.2

B. Charlier et al. / Chemical Geology 234 (2006) 264–290

8V-9S 298 LCZ 30.98 18.17 14.01 21.25 0.13 3.80 5.30 4.21 0.55 0.32 98.72 0.152 47.0 43.8 3.7 0.0 1.1 4.4
12V-13N 110 LCZ 35.06 15.04 14.68 19.19 0.13 4.49 5.62 3.73 0.61 0.35 98.90 0.190 52.8 36.0 5.4 0.0 1.0 4.8
12V-3S 230 LCZ 30.70 18.33 13.27 21.97 0.13 5.10 4.81 3.69 0.47 0.21 98.68 0.188 43.0 48.3 6.3 0.0 0.0 2.3
12V-3S 255 LCZ 30.60 18.72 14.00 21.61 0.13 3.86 5.11 3.97 0.49 0.22 98.71 0.152 44.0 48.9 4.9 0.0 0.2 2.0
12V-3S 285 LCZ 31.46 17.84 14.56 20.53 0.12 3.54 5.37 4.10 0.52 0.23 98.27 0.147 49.1 45.2 3.5 0.0 0.0 2.2
12V-10S 110 LCZ 31.54 17.89 14.05 21.01 0.13 4.05 5.22 3.92 0.55 0.28 98.64 0.162 45.0 45.6 4.8 0.0 0.2 4.4
12V-14S 75 LCZ 31.84 17.48 12.37 22.21 0.15 5.65 4.88 3.51 0.56 0.37 99.02 0.203 37.9 48.2 7.7 0.0 0.7 5.4
12.5V-6S 111 LCZ 30.98 18.36 13.99 21.11 0.12 4.03 5.12 3.68 0.49 0.23 98.11 0.160 44.0 47.0 5.4 0.0 0.3 3.3
12.5V-6S 153 LCZ 30.91 18.54 13.62 21.20 0.12 3.60 5.27 4.09 0.51 0.24 98.10 0.145 50.1 43.5 3.4 0.0 0.4 2.6
12.5V-6S 180 LCZ 29.85 19.30 13.83 21.92 0.13 3.69 5.15 3.90 0.50 0.27 98.54 0.144 48.0 45.4 2.1 0.0 0.4 4.0
12.5V-6S 210 LCZ 33.17 16.53 15.03 19.54 0.12 3.66 5.61 4.15 0.62 0.33 98.76 0.158 51.2 41.5 4.8 0.0 0.7 1.8
12.5V-6S 239 LCZ 32.87 16.73 14.23 20.50 0.13 4.02 5.37 4.09 0.62 0.37 98.93 0.164 47.0 45.2 3.7 0.0 0.0 4.2
16V-3S 166 LCZ 31.35 18.09 13.79 21.31 0.13 4.48 5.14 3.87 0.52 0.25 98.93 0.174 45.7 45.2 4.7 0.0 0.2 4.2
16V-3S 208 LCZ 30.35 18.87 13.64 22.09 0.13 3.62 5.04 3.95 0.56 0.30 98.55 0.141 50.4 42.0 3.5 0.0 0.4 3.6
16V-8L 5 LCZ 31.80 17.52 13.64 21.41 0.14 4.57 5.06 3.58 0.60 0.34 98.66 0.176 43.7 43.1 7.4 0.0 0.2 5.6
16V-8L 30 LCZ 33.81 16.10 14.75 19.61 0.12 3.65 5.59 3.94 0.68 0.37 98.62 0.157 48.2 42.4 4.2 0.0 1.2 4.0
16V-8L 51 LCZ 37.26 13.43 15.50 17.89 0.12 4.03 5.89 4.01 0.63 0.31 99.06 0.184 49.0 38.4 9.0 0.0 0.5 3.1
16V-8L 68 LCZ 34.54 15.22 13.09 20.85 0.15 5.15 5.49 3.45 0.56 0.31 98.81 0.198 36.6 42.0 8.6 0.0 0.8 11.5
Average of LCZ 31.60 17.80 13.89 21.18 0.13 4.14 5.21 3.86 0.54 0.29 98.64 0.164 46.2 44.7 4.9 0.0 0.4 3.8
8V-2S 210 LMZ 33.96 15.91 14.57 19.79 0.13 3.74 5.66 4.08 0.67 0.40 98.91 0.159 50.8 40.0 3.3 0.0 1.4 4.5
8V-7S 140 LMZ 31.85 17.44 13.89 21.14 0.13 3.92 5.31 3.76 0.62 0.43 98.49 0.156 47.5 40.3 7.9 0.0 0.1 4.2
8V-8S 90 LMZ 33.75 15.98 13.50 20.75 0.14 4.28 5.46 3.68 0.76 0.58 98.88 0.171 50.0 38.9 1.6 0.0 0.2 9.3
8V-8S 120 LMZ 36.09 14.11 14.93 18.56 0.13 3.99 5.76 4.18 0.83 0.49 99.07 0.177 47.8 38.3 8.9 0.0 0.0 5.0
12V-13N 140 LMZ 41.04 9.33 17.12 15.87 0.11 3.91 6.27 4.23 0.89 0.39 99.16 0.198 62.5 25.5 4.4 0.0 0.5 2.7
12V-14S 47 LMZ 39.18 11.78 16.07 16.71 0.12 3.35 6.07 4.50 0.95 0.49 99.22 0.167 58.1 32.3 4.1 0.0 2.5 3.1
12V-14S 96 LMZ 40.29 10.67 15.57 16.40 0.13 4.65 6.19 4.07 0.87 0.51 99.35 0.221 52.7 34.8 6.8 0.0 0.7 4.5
12.5V-6S 264 LMZ 46.84 5.39 16.10 13.07 0.14 5.18 7.15 4.10 1.07 0.69 99.73 0.284 55.5 15.1 12.6 0.0 9.8 5.0
16V-6N 57 LMZ 36.16 13.59 14.35 18.80 0.13 5.01 5.61 3.90 0.67 0.48 98.70 0.210 48.1 34.2 9.8 0.0 2.3 5.0
16V-3S 231 LMZ 33.17 16.59 14.44 20.13 0.13 3.57 5.49 4.11 0.66 0.40 98.69 0.151 49.1 39.8 5.1 0.0 1.6 4.4
16V-4S 103 LMZ 42.53 9.00 16.66 14.28 0.11 4.89 6.42 4.08 0.78 0.36 99.11 0.255 57.3 29.6 5.0 0.0 2.6 5.1
16V-8L 73 LMZ 40.85 10.01 16.41 15.68 0.11 4.19 6.27 4.14 0.88 0.42 98.96 0.211 52.9 33.0 5.3 0.0 4.3 4.5
16V-8L 74 LMZ 33.94 15.03 12.27 23.01 0.18 5.14 5.15 3.09 0.53 0.33 98.67 0.183 39.8 42.9 7.8 0.0 5.0 3.8
Average of LMZ 37.67 12.68 15.07 18.01 0.13 4.29 5.91 3.99 0.78 0.46 99.00 0.196 51.7 34.2 6.4 0.0 2.4 4.7
8V-4N 210 UCZ 32.16 17.16 12.89 21.69 0.14 6.26 4.77 3.48 0.46 0.23 99.24 0.224 41.8 42.0 7.9 2.0 2.1 3.2
8V-4N 238 UCZ 29.76 19.04 12.58 23.11 0.14 5.50 4.59 3.53 0.44 0.23 98.92 0.192 45.8 43.1 4.8 2.5 0.1 3.2
8V-5N 77 UCZ 32.50 17.06 11.12 23.17 0.17 7.74 4.20 2.97 0.42 0.21 99.56 0.250 33.7 39.0 18.3 1.8 0.9 2.6
8V-5N 113 UCZ 31.87 17.47 11.62 22.87 0.16 7.21 4.30 3.14 0.49 0.25 99.38 0.240 29.9 42.6 20.1 1.2 0.9 4.0
8V-6N 17 UCZ 33.55 16.79 10.42 23.77 0.18 7.10 3.99 2.59 0.62 0.33 99.34 0.230 24.5 38.5 27.0 1.1 0.1 4.3
8V-6N 40 UCZ 35.51 15.09 11.91 21.34 0.16 6.20 4.62 2.98 0.76 0.41 98.98 0.225 35.6 39.8 18.4 0.5 1.2 4.6
8V-2S 16 UCZ 34.61 14.88 8.65 24.89 0.21 10.11 3.57 2.18 0.53 0.35 99.98 0.289 24.5 38.5 27.8 0.3 0.1 4.3
8V-2S 65 UCZ 34.31 16.08 11.19 22.33 0.15 6.95 4.27 3.02 0.65 0.35 99.30 0.237 37.8 36.7 20.8 0.4 0.0 4.4
8V-2S 120 UCZ 34.80 15.60 12.36 21.17 0.15 6.18 4.65 3.41 0.66 0.33 99.28 0.226 37.3 39.6 18.3 0.4 0.5 3.3
8V-2S 158 UCZ 33.23 16.37 12.94 21.39 0.14 5.91 4.82 3.55 0.58 0.35 99.28 0.216 43.2 39.2 8.3 2.8 0.2 4.7
8V-9S 44 UCZ 35.61 15.01 11.88 21.50 0.17 7.23 4.36 3.14 0.47 0.18 99.55 0.252 37.7 36.8 22.8 0.0 0.1 2.6
8V-9S 70 UCZ 35.30 14.86 13.57 19.89 0.14 6.40 4.95 3.50 0.51 0.22 99.34 0.243 39.2 41.3 13.6 0.0 0.2 4.0
8V-9S 100 UCZ 29.07 19.61 11.92 23.75 0.15 6.01 4.35 3.30 0.42 0.21 98.78 0.202 41.8 41.3 7.9 3.1 0.1 5.2

B. Charlier et al. / Chemical Geology 234 (2006) 264–290

8V-9S 130 UCZ 29.37 19.27 12.41 23.27 0.14 5.75 4.53 3.69 0.43 0.22 99.08 0.198 40.9 45.9 6.9 2.3 0.0 3.6
8.2V-9S 34 UCZ 29.36 19.43 9.10 26.09 0.19 8.56 3.56 2.24 0.35 0.26 99.14 0.247 28.9 48.3 16.4 3.3 0.5 2.5
8.2V-9S 69 UCZ 31.97 17.07 12.20 22.42 0.15 7.01 4.51 3.04 0.46 0.22 99.05 0.238 38.2 43.2 10.1 4.1 0.4 3.1
11.5V-5S 97 UCZ 29.92 18.88 12.32 22.97 0.14 5.79 4.51 3.18 0.44 0.24 98.39 0.201 44.7 41.5 5.4 3.0 0.1 4.5
11.5V-5S 124 UCZ 30.48 18.60 13.17 21.93 0.13 5.02 4.95 3.45 0.46 0.23 98.42 0.186 45.6 44.0 7.1 0.2 0.0 3.0
12V-13N 10 UCZ 33.71 15.62 12.81 21.23 0.15 6.88 4.76 3.41 0.50 0.24 99.31 0.245 40.4 38.5 13.7 2.6 0.1 3.1
12V-13N 34 UCZ 31.34 17.47 12.02 22.92 0.15 6.84 4.44 3.09 0.45 0.25 98.97 0.230 37.8 40.6 10.2 6.3 1.3 2.9
12V-13N 59 UCZ 32.71 16.59 12.78 21.56 0.14 6.70 4.74 3.24 0.51 0.26 99.22 0.237 44.4 40.4 4.4 6.0 0.2 3.9
12V-3S 29 UCZ 35.24 14.42 12.65 20.72 0.15 7.74 4.66 3.26 0.53 0.23 99.60 0.272 40.7 36.5 13.7 3.6 0.3 4.6
12V-3S 54 UCZ 29.46 19.17 10.28 25.05 0.17 8.32 3.76 2.74 0.35 0.16 99.46 0.249 35.7 43.0 12.5 5.1 0.0 2.2
12V-3S 87 UCZ 29.62 18.94 10.70 24.67 0.16 7.81 3.91 2.88 0.38 0.19 99.26 0.240 35.5 46.4 7.5 5.3 0.2 4.2
12V-3S 120 UCZ 32.28 16.93 12.10 22.26 0.15 7.19 4.49 3.29 0.49 0.24 99.42 0.244 39.4 41.4 10.8 3.3 0.4 3.7
12V-3S 154 UCZ 31.17 17.79 11.83 23.00 0.15 6.98 4.38 3.24 0.47 0.24 99.25 0.233 40.6 40.4 10.0 3.5 0.4 4.5
12V-3S 194 UCZ 30.90 18.13 12.29 22.75 0.15 6.40 4.53 3.38 0.44 0.22 99.19 0.220 40.7 41.8 10.1 2.0 0.5 3.2
12V-9S 45 UCZ 38.20 11.20 14.45 19.77 0.13 6.14 5.50 3.31 0.61 0.22 99.54 0.237 42.5 34.4 13.5 2.1 1.5 5.3
12V-9S 77 UCZ 31.57 17.39 12.59 22.55 0.15 6.26 4.71 3.38 0.45 0.20 99.25 0.217 42.7 41.2 8.3 3.9 0.4 1.9
12V-10S 70 UCZ 37.25 12.32 14.76 18.71 0.12 6.16 5.48 3.74 0.57 0.23 99.34 0.248 47.8 31.5 8.9 3.4 1.1 3.9
12V-7L 10 UCZ 30.63 18.19 11.07 23.72 0.16 7.52 4.09 2.96 0.43 0.21 98.98 0.241 34.9 42.8 12.8 5.2 0.3 2.4
12V-7L 47 UCZ 34.50 15.41 13.82 20.07 0.14 4.91 5.26 3.88 0.67 0.39 99.05 0.197 48.3 35.7 10.0 0.3 0.8 4.6
12V-8L 3 UCZ 30.34 18.53 11.15 23.98 0.16 7.49 4.10 2.98 0.40 0.20 99.33 0.238 35.9 45.4 13.0 2.7 0.2 2.1
12V-8L 40 UCZ 32.24 16.95 12.92 21.65 0.14 5.79 4.82 3.37 0.53 0.30 98.71 0.211 44.5 38.8 7.0 2.4 2.2 4.0
12.5V-6S 47 UCZ 31.26 17.87 12.45 22.50 0.15 6.23 4.61 3.18 0.46 0.24 98.95 0.217 35.7 50.2 7.6 3.4 0.7 2.5
12.5V-6S 79 UCZ 29.07 19.61 12.37 23.56 0.14 5.70 4.47 3.20 0.41 0.20 98.73 0.195 40.7 47.9 6.3 1.3 0.6 2.6
12.5V-2L 5 UCZ 30.56 18.08 10.58 24.34 0.16 8.49 3.88 2.75 0.40 0.20 99.44 0.259 30.4 46.9 13.3 6.0 1.0 2.1
13V-8L 5 UCZ 36.64 13.36 13.56 19.37 0.14 7.16 5.07 3.51 0.53 0.28 99.62 0.270 45.7 34.9 12.1 2.3 0.7 4.2
16V-6N 15 UCZ 33.01 15.94 12.72 21.78 0.15 6.53 4.82 3.46 0.52 0.32 99.25 0.231 44.0 39.7 8.6 3.0 0.4 2.6
16V-3S 7 UCZ 32.14 16.73 11.94 22.47 0.15 7.21 4.41 3.18 0.47 0.22 98.92 0.243 31.8 42.8 17.7 4.2 0.1 2.8
16V-3S 47 UCZ 32.24 16.83 12.34 22.13 0.15 6.99 4.59 3.29 0.47 0.23 99.26 0.240 38.8 41.1 12.4 3.3 0.3 3.2
16V-3S 84 UCZ 30.96 17.74 12.08 22.94 0.15 6.95 4.43 3.26 0.42 0.21 99.14 0.233 42.2 40.1 8.9 3.5 0.4 3.5
16V-3S 127 UCZ 30.91 18.16 12.76 22.34 0.14 5.91 4.68 3.51 0.46 0.22 99.09 0.209 45.7 43.0 5.5 1.9 0.1 3.2
16V-4S 6 UCZ 32.82 16.22 12.42 21.89 0.15 7.05 4.64 3.32 0.50 0.23 99.24 0.244 37.3 42.2 8.8 5.9 1.3 2.1

273
(continued on next page)
274 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
Biotite
4.0 5.2. Modal proportions and spatial variations in
2.4
3.5
3.0
3.1
6.5
2.5
4.8
8.5
1.4
2.8
3.5
5.8
4.2
cumulus mineralogy
Cpx
0.5
0.9
0.5
3.2
2.1
3.3
1.3
1.5
0.9
1.0
2.3
4.4
3.9
2.4
Our petrographical study confirms the variations of
modal proportions previously described by Gierth and
Olivine

Krause (1973) and reveals systematic variations in the

3.2
1.7
2.7
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
cumulus assemblage throughout the deposit. The zoning
is observed in each of the three sections studied (800,
10.2

18.2
15.3
10.9
18.5
10.1

12.5

13.8

12.4
8.6
11.9

7.4

11.5

5.7
Opx

1200 and 1600), suggesting that variations along the

length of the ore body are not significant.
47.7
43.2
41.3
32.9
28.1
31.1
28.9
28.4
27.9
32.8
30.7
26.5
32.7
30.0
Ilm

Based on the modal and textural variations, the four

zones identified by Kullerud (2003) on the basis of
Plagio
33.4
41.1
38.9
40.3
50.4
47.0
48.8
54.6
55.2
51.7
52.7
51.5
50.9
50.3

mining whole-rock data have been defined in the three

sections (Fig. 6): (1) the Lower Central Zone (LCZ);
0.232
0.202
0.231
0.219
0.248
0.220
0.254
0.247
0.240
0.251
0.247
0.240
0.222
0.239

(2) the Upper Central Zone (UCZ); (3) the Upper

Mg#

Marginal Zone (UMZ) and (4) the Lower Marginal

Zone (LMZ). As detailed later, the Tellnes ilmenite
99.21
98.96
99.19
99.41
99.53
99.00
99.14
99.43
98.98
99.47
98.94
99.41
99.59
99.29
Total

deposit is interpreted as a sill-like body that has

subsided so that it has now a synclinal shape. The
P2O5
0.20
0.24
0.25
0.50
0.31
0.45
0.26
0.38
0.38
0.40
0.23
0.46
0.57
0.39

original morphology and spatial distribution of the

different zones has been therefore strongly modified by
K2O
0.33
0.47
0.49
0.76
0.55
0.85
0.72
0.82
0.78
0.81
0.66
0.93
1.14
0.80

later deformation.
The Lower Central Zone (LCZ) represents rocks from
the lower part of the ore body and also the upper south part
Na2O
2.48
3.44
3.20
3.95
3.58
3.80
3.34
4.02
4.01
3.69
3.93
4.06
4.65
3.90

in sections 800 and 1200. In the LCZ, plagioclase and

ilmenite constitute ∼90% of the mode and are the only
CaO
3.31
4.64
4.49
5.45
5.37
5.69
5.45
5.94
6.09
6.05
6.02
6.22
5.79
5.81

cumulus phases (pi-C). Their average amounts are 46.2

and 44.7 wt.% respectively. It is the most ilmenite-rich
8.18
5.79
6.79
5.36
6.33
4.95
6.12
5.36
5.06
5.94
5.25
4.93
5.09
5.44
MgO

part of the deposit. Orthopyroxene is also present but in

minor amounts as small interstitial grains. Olivine and
MnO

large grains of magnetite are absent. Magnetite is only

0.18
0.15
0.15
0.14
0.14
0.13
0.13
0.13
0.12
0.14
0.12
0.12
0.14
0.13

present as rare small grains associated with sulphides.

Other intercumulus minerals are biotite, apatite, clinopyr-
FeOtot
27.04
22.83
22.51
19.14
19.17
17.53
17.94
16.30
16.01
17.69
15.97
15.65
17.82
17.32

oxene and locally amphibole.

The Upper Central Zone (UCZ) comprises the upper
Al2O3

12.37
12.07
14.06
13.92
14.83
14.43
15.20
15.47
14.41
15.80
15.60
14.38
14.81
9.11

portion of the intrusion, excluding the marginal rocks.

Plagioclase, ilmenite, orthopyroxene and olivine are the
cumulus minerals (piho-C) with average amounts of
9.89

9.69

11.02
21.66
18.23
17.04
13.26
12.85
12.17

10.43
10.33
10.17
10.42

10.95
TiO2

respectively 38.9, 41.3, 11.9 and 2.7 wt.%. The

orthopyroxene content increases upwards through the
26.72
30.80
32.21
36.79
37.31
38.60
40.88
40.85
40.73
40.17
40.54
41.75
39.06
39.67
SiO2

UCZ. Intercumulus magnetite is invariably present.

Clinopyroxene, biotite, apatite and locally amphibole
are also intercumulus phases. The boundary between the
UMZ
UMZ
UMZ
UMZ
UMZ
UMZ
UMZ
UMZ
UMZ
UMZ
Zone
UCZ
UCZ

UCZ and LCZ is well defined as olivine and large crystals

of prismatic orthopyroxene appear together. This bound-
Depth

ary has a trough-shape and is grossly parallel in section to

25
65

67

46
71
80
82
7

11
156

104

the bottom of the ore body.

Table 3 (continued)

The Upper Marginal Zone (UMZ) is situated at the

Average of UMZ
Average of UCZ

margins of the ore body between the UCZ and the

contact with the host anorthosite. It is characterized by a
Drill core

7.8V-1N

8.2V-9S
12V-2N
12V-7N
12V-7L
12V-8L
12V-8L
16V-4S
16V-4S

00-30C
8V-4N
8V-6N

higher abundance of plagioclase (50.3 wt.% on average)

and apatite, and a lower content of ilmenite (30.0 wt.%
 B. Charlier et al. / Chemical Geology 234 (2006) 264–290 275

Fig. 5. Microphotographs showing the main textures observed in the Tellnes ilmenite deposit. A. Large prismatic orthopyroxene with Schiller
exsolutions, with ilmenite inclusions. Larger ilmenite grains have interstitial morphologies, wrapping orthopyroxene and plagioclase laths. Traces of
biotite are common (sample 8V-4N, 156; UCZ); B. Small anhedral orthopyroxene interstitial to plagioclase. Note lobate contacts between plagioclase
and ilmenite (sample 12V-3S, 255; LCZ); C. Symplectitic intergrowths of orthopyroxene and oxides replacing primary olivine (sample T2 from Diot
et al., 2003; UMZ); D. Fine-grained jotunite with the two pyroxenes in similar proportions, antiperthitic plagioclase, ilmenite and relatively large
proportion of apatite (sample 12V-10S, 64; UMZ); E: Poikilitic texture with large plagioclase and pyroxene oikocrysts enclosing numerous small
ilmenite chadacrysts giving a “Dalmatian texture” (sample 99-24 b, NE margin of the ore body; UMZ); F: same view under crossed nicols.

on average) than the UCZ. Magnetite is present and and ilmenite/magnetite are common and resulted from
clinopyroxene forms relatively abundant subhedral the oxidation of cumulus olivine. The thickness of the
grains. Symplectitic intergrowths of orthopyroxene UMZ is about 15 to 20 m and the transition to the UCZ
276 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
Fig. 6. (A–C) Cross sections (800, 1200 and 1600) through the Tellnes
ilmenite deposit showing the spatial distribution of different zones and
plagioclase compositions (An and Or content) in selected samples.
LCZ = Lower Central Zone; UCZ = Upper Central Zone; UMZ = Upper
Marginal Zone; LMZ = Lower Marginal Zone; p = plagioclase; i = il-
menite; h = orthopyroxene; o = olivine; -C = cumulate. Cumulate no-
menclature after Irvine (1982).
is gradational and characterized by a decrease in the
proportion of plagioclase to ilmenite. The UMZ is
absent in section 1600 because the margins of the ore
body are only in contact with the LCZ.
The Lower Marginal Zone (LMZ) is situated between
the LCZ and the contact. It is similar to the UMZ:
relatively plagioclase-rich and ilmenite-poor (51.2 and
34.4 wt.% respectively). However, symplectitic inter-
growths are absent (due to the absence of olivine in the
primary assemblage) and orthopyroxene is less abundant.
As with the UMZ, apatite is more abundant than in the
LCZ.
6. Analytical results
6.1. Plagioclase compositions
Plagioclase compositions are widely used to charac-
terize stratigraphic variations (cryptic layering) in
cumulates in layered intrusions in order to identify
magma chamber processes such as fractional crystalli-
zation, magma recharge, assimilation or trapped liquid
crystallization. Even though plagioclase is known to
retain its primary composition as a result of slow NaSi–
CaAl exchange (Grove et al., 1981), reequilibration and
diffusion could nevertheless be significant in deep-
seated plutonic rocks. Deformed twinning and local
polygonization are common in Tellnes though less
extensive than in other nearby intrusions such as
Bjerkreim-Sokndal and the anorthosite massifs. This
weak granulation is possibly due to the relatively ductile
behaviour of ilmenite which has absorbed most of the
strain. Nevertheless, recrystallization of primary euhe-
dral laths of plagioclase might have overprinted and
blurred any original plagioclase zoning. In order to
eliminate the variations resulting from recrystallization
of zoned crystals, we have characterized the plagioclase
by its bulk composition rather than by using a
microprobe.
Previous data on plagioclase from Wilmart et al.
(1989) described it as having a composition in the range
An45–39. Compositions An45–42 were reported in the
centre of the ore body and, more sodic plagioclase
(down to An39) occurred in some apophyses and in
plagioclase-rich rocks at the margins. The new data
reveal a larger range of compositions than previously
described (Table 1). Bulk plagioclase compositions vary
from An48.8 to An39.3 (An = 100 [Ca/Ca + Na]) and Or2.8
to Or7.1 (Or = 100 [K/Ca + Na + K]). Iron in plagioclase
has been shown to be a monitor of magma composition
(e.g. Tegner, 1997) and an indicator of the oxygen
fugacity (Phinney, 1992). However, our analyses on
mineral separates do not allow us to make detailed
investigation of the iron content of plagioclase as it
contains numerous small oxide inclusions. MgO, TiO2-
and MnO are minor components that are very sensitive
 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
to contamination and their variations will not be
considered.
Spatial variations of An and Or content in plagioclase
are shown in Fig. 6. Compositions are relatively
constant in the LCZ and UCZ (∼ An49–45). The major
variation in plagioclase composition is observed at the
margins of the ore body where an abrupt decrease in An
(An43–39) and increase in Or (Or4–7) is commonly
observed.
Sr and Ba (Table 1) have also been analysed in
plagioclase separates. Sr is compatible in plagioclase
(see Vander Auwera et al., 2000 for a review) whereas
Ba is incompatible (Bindeman et al., 1998). The range
of variation is quite small for Sr (964 to 1182 ppm) and
more significant for Ba (154 to 355 ppm). The Sr
content is higher in samples from LCZ and UCZ
whereas samples from LMZ and UMZ are compara-
tively enriched in Ba. Binary diagrams (Fig. 7) show
that An and Sr (compatible behaviour) are positively
correlated. Or and Ba (incompatible behaviour) are
correlated positively and negatively with An and Sr.
These trends do not result from variations of partition
coefficients for Sr and Ba because they increase linearly
with decreasing An content (Bindeman et al., 1998).
This is evident from the An–Sr relation which is
opposite to the trend predicted by variations in partition
coefficients. On the other hand, it is clear that these
Fig. 7. Binary diagrams for plagioclase composition from the Tellnes ilmenite deposit. Symbols are related to a range of P2O5 contents in the bulk
cumulate.
277
trends are closely related to the P2O5 content in the bulk
cumulate. P2O5-richer cumulates have plagioclase with
higher Or and Ba contents while P2O5-poorer samples
have plagioclase with higher An and Sr contents.
6.2. Sr isotope geochemistry
Sr isotopic ratios and Rb and Sr concentrations have
been analysed for 21 separated plagioclases (Table 2).
Measured 87/86Sr ratios display a very restricted range
(0.704396–0.704693). These values are slightly lower
than data reported by Wilmart et al. (1989) on four bulk
cumulates from the ore body (0.70487–0.70510). They
correspond to the lowest values in the Rogaland
Anorthosite Province (see Bolle et al., 2003, for a
review). The spatial variability of 87/86Sr(920) is shown in
Fig. 8. Systematic variations of this ratio are observed in
section 1200 (Fig. 8A). The lowest 87/86Sr(920) ratio
(0.704368) is found at the base of the section and it
increases continuously towards the top, where it reaches
0.704519. Sample 12.5V-6S (264), situated 10 cm from
the contact with the host anorthosite, displays the
highest ratio (0.704533). Values for 87/86 Sr(920) in
section 1600 (Fig. 8B) are in the same range as values
in the upper part of section 1200. The range of variation
is minor, unsystematic, and probably not genetically
significant.
278 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
Fig. 8. Spatial distribution of 87/86Sr(920) in plagioclase from cross
section 1200 and 1600 and location of the LCZ and UCZ in the Tellnes
ilmenite deposit. Only the three last decimals of the ratios are
indicated.
6.3. Whole-rock compositions
Ninety seven new whole-rock analyses (Table 3) are
plotted in major element binary diagrams (Fig. 9)
together with mining data. SiO2 and TiO2 have been
chosen as variation indices, as SiO2 is essentially
correlated with the modal abundance of plagioclase
and orthopyroxene and TiO2 reflects the amount of
ilmenite. The distribution of the new analyses mimics
the variation of mining data, except for some plagio-
clase-rich samples from the mining database. These
plagioclase-rich samples consist of mixtures of the ore
and the host anorthosite or anorthositic xenoliths and are
thus not representative of rocks from the Tellnes
ilmenite deposit.
Compositional ranges of cumulus minerals (plagio-
clase, ilmenite, orthopyroxene and olivine) are repre-
sented by a circle or indicated by an arrow. One of the
rock samples from the LMZ (jot) has been circled. This
sample 12.5V-6S (264) has the lowest TiO2 and FeOtot
and the highest SiO2 and CaO of our suite of samples. It
is also characterized by the highest P2O5 (0.69%, Table
3). The texture (Fig. 5D) and composition of this sample
are similar to jotunites described from elsewhere in the
Rogaland Anorthosite Province (Table 4). These
jotunites (hypersthene monzodiorites rich in FeO and
TiO2 and relatively low in CaO) have been called
“primitive jotunites” by Vander Auwera et al. (1998)
because they are the least differentiated jotunites
observed in Rogaland. Similar compositions are referred
to as ferrodiorites or monzonorites by other authors (e.g.
Mitchell et al., 1996) and are commonly associated with
massif-type anorthosites. In the Rogaland Anorthosite
Province, these rocks are found as dykes (Duchesne et
al., 1974, 1985, 1989), as chilled margins of plutons
such as Hidra and the Bjerkreim-Sokndal layered
intrusion (Demaiffe and Hertogen, 1981; Duchesne
and Hertogen, 1988; Robins et al., 1997), or as pillows
in the Apophysis (Bolle, 1996). Sample 12.5V-6S (264)
is considered to represent a melt composition.
Compositional arrays are distributed within triangles
or display well-defined linear trends. Samples from LCZ
and UCZ are well discriminated by the MgO content
which is significantly higher in UCZ. This is a
consequence of the presence of cumulus orthopyroxene
and olivine in UCZ. Samples from LMZ and UMZ are
relatively TiO2-poor and are closer to the jotunite
composition. These diagrams clearly demonstrate that
compositional variations in the rocks essentially result
from the occurrence of cumulus orthopyroxene and
olivine in the UCZ (in addition to plagioclase and
ilmenite in the LCZ), and from the content of a trapped
jotunitic liquid component that is more abundant in the
marginal zones.
7. Discussion
7.1. Sequence of crystallization: comparison with the
Bjerkreim-Sokndal layered intrusion
The Bjerkreim-Sokndal layered intrusion crops out in
the central part of the Rogaland Anorthosite Province
(Michot, 1965; Duchesne, 1987; Wilson et al., 1996). The
lower part of this intrusion consists of a 7000-m thick
series of cumulates (the “Layered Series”) and the upper
part comprises acidic rocks, many of which are interpreted
to represent liquid compositions (Duchesne and Wilmart,
1997). All units of the AMC series are developed:
anorthosite, troctolite, leuconorite, norite, gabbronorite,
jotunite, mangerite, quartz mangerite and charnockite.
The parental magma of this intrusion has a jotunitic
composition (Duchesne and Hertogen, 1988; Vander
 B. Charlier et al. / Chemical Geology 234 (2006) 264–290 279

Fig. 9. Binary major elements variation diagrams of bulk cumulates from the Tellnes ilmenite deposit. The compositional ranges of cumulus minerals
(plag = plagioclase, ilm = ilmenite, opx = orthopyroxene, ol = olivine) is represented by an elliptic area or indicated by an arrow (data from this study
and Charlier et al., submitted for publication). The circled triangle is sample 12.5V-6S (264) (jot = jotunite). Other jotunites are from Table 4.
Calculated adcumulate poles pi-C and piho-C are represented by a circle.
280 B. Charlier et al. / Chemical Geology 234 (2006) 264–290

Table 4
Composition of the marginal sample from the Tellnes deposit compared to selected primitive jotunites from the Rogaland Anorthosite Province
Sample 12.5V-6S (264) 80.12.3a 80.12.4 80.12.1 91.14.1 70.20 200.2/2 72.34 HLCA TJ
Reference This study (4) (3) (3) (5) (1) (2) (1) (6) (6)
Locality Tellnes Tjörn-BKSK BKSK BKSK BKSK Hidra Hidra Hidra
SiO2 46.84 49.39 49.27 49.57 50.38 48.00 48.68 49.53 50.0 49.5
TiO2 5.39 3.67 5.22 4.95 4.62 4.61 4.12 3.82 1.85 3.46
Al2O3 16.10 15.81 14.00 14.11 14.65 14.20 12.40 14.50 17.5 16.0
FeOtot 13.07 13.11 14.31 13.55 13.20 14.90 16.62 13.68 10.8 13.1
MnO 0.14 – 0.15 0.14 – 0.18 – – 0.15 0.13
MgO 5.18 4.54 5.71 5.46 4.98 4.60 5.96 4.90 6.67 4.54
CaO 7.15 6.87 6.18 6.21 7.01 6.41 6.04 6.00 8.78 6.82
Na2O 4.10 3.50 3.72 3.57 3.53 3.65 3.24 3.50 2.93 3.65
K2O 1.07 0.96 1.06 1.00 1.07 1.08 1.32 1.95 0.44 0.94
P2O5 0.69 0.71 0.79 1.07 0.73 0.81 0.40 0.91 0.16 0.63
Total 99.73 98.56 100.41 99.63 100.17 98.44 98.78 98.79 99.3 98.8
FeOa 11.40 11.51 12.57 11.91 11.55 13.10 14.65 11.99 9.43 11.49
Fe2O3a 1.85 1.78 1.94 1.82 1.83 2.00 2.19 1.88 1.52 1.79
mg# 0.414 0.382 0.416 0.418 0.402 0.355 0.390 0.390 0.524 0.382
NAb 0.562 0.509 0.576 0.558 0.537 0.562 0.556 0.515 0.423 0.522
NOr 0.096 0.092 0.108 0.103 0.107 0.109 0.149 0.189 0.042 0.088
References: (1) Duchesne et al. (1974); (2) Demaiffe (1977); (3) Duchesne et al. (1989); (4) Duchesne and Hertogen (1988); (5) Vander Auwera et al.
(1998); (6) Longhi et al. (1999).
a
Calculated from Kress and Carmichael (1991) at FMQ and 1100 °C.

Auwera and Longhi, 1994; Robins et al., 1997). Detailed
mapping and studies of mineral compositions (Michot,
1965; Duchesne, 1972; Nielsen and Wilson, 1991; Jensen
et al., 1993; Nielsen et al., 1996; Jensen et al., 2003;
Wilson and Overgaard, 2005) as well as experimental
studies on the jotunitic parental magma (Vander Auwera
and Longhi, 1994) have resulted in a detailed picture of
the sequence of crystallization and magma chamber
processes (Fig. 10). The Layered Series has been
subdivided into 6 megacyclic units (MCU 0, IA, IB, II,
III, and IV). Using the observed regressions in mineral
compositions (Duchesne, 1972; Nielsen and Wilson,
1991; Jensen et al., 1993) and abrupt variations in isotopic
ratios (Nielsen et al., 1996; Barling et al., 2000), it has
been shown that each MCU resulted from the crystalli-
zation of a new magma influx that mixed with residual
resident liquid. Isotopic variations also provide evidence
for continuous assimilation of surrounding gneissic rocks
(Tegner et al., 2005). The sequence of crystallization in the
Layered Series (Wilson et al., 1996) is usually: plagio-
clase, ilmenite, (or plagioclase + olivine + ilmenite
+ magnetite), orthopyroxene, magnetite, clinopyroxene
together with apatite, and then pigeonite (instead of
orthopyroxene). Fayalitic olivine reappears in the Tran-
sition Zone, followed by mesoperthite in mangerites and
finally quartz appears in quartz mangerites.
In the Tellnes deposit, the LCZ contains the most
primitive cumulates and has plagioclase and ilmenite as
cumulus minerals (pi-C). The LCZ would thus corre-
spond to pi-C of zone a in Bjerkreim-Sokndal. This
suggests that parental magmas to both intrusions were
related as they had the same liquidus minerals. Cumulus
orthopyroxene and olivine appear in the UCZ. This
sequence of crystallization differs from the Bjerkreim-
Sokndal layered intrusion where orthopyroxene does
not seem to have been in equilibrium with olivine. As
discussed below, these dissimilarities can be easily
explained by small differences in parental magma
compositions. Apatite and clinopyroxene do not appear
as cumulus minerals in Tellnes. Duchesne and Charlier
(2005) have shown that apatite appeared in Bjerkreim-
Sokndal as a liquidus phase when the proportion of
residual liquid (F) was about 0.6. This implies that the
Tellnes deposit corresponds to the early stage of
evolution of a jotunitic magma and is only the lower
part of a larger intrusion. More evolved cumulates are
not observed at the present level of exposure.
Toward the base of the Bjerkreim-Sokndal layered
intrusion, a thick sequence of anorthositic cumulates (p-C)
crystallized before pi-C, but such cumulates are absent in
Tellnes. The origin of these anorthosites in MCU Ia and Ib
has been discussed by Duchesne and Charlier (2005), who
concluded that the parental magma of these units could
have been different from the parental magma of MCU II to
IV. Moreover, a higher TiO2 content in the Tellnes parental
magma could have been responsible for the earlier
saturation in ilmenite (Toplis and Carroll, 1995), prevent-
ing the crystallization of a large volume of anorthosite.
 B. Charlier et al. / Chemical Geology 234 (2006) 264–290 281

Fig. 10. Cumulate stratigraphy of the Bjerkreim-Sokndal layered intrusion as developed in the northern part of the intrusion along the axial trace of the
syncline. The intrusion is subdivided into 6 megacyclic units (MCU 0, IA, IB, II, III and IV) in which compositional zones (a to f) are recognized. The
compositional ranges of cumulus minerals are indicated (after Wilson et al., 1996). Cumulus assemblages following the nomenclature of Irvine (1982)
in zones a to f : a = p(± i)-C; b = pimo-C; c = pih-C; d = pihm-C; e = pihmac-C; f = pih’mac-C. Mineral abbreviations: p = plagioclase; h = orthopyroxene;
h′ = inverted pigeonite; o = olivine; i = ilmenite; m = magnetite; a = apatite; c = clinopyroxene; -C = cumulus.

Finally, the much higher proportion of ilmenite in
Tellnes compared to Bjerkreim-Sokndal, where TiO2 in
bulk cumulates is around 3.7% on average (Duchesne
and Charlier, 2005), necessitates either sorting and
accumulation of ilmenite in the ore body or a
significantly higher cotectic proportion of ilmenite in
Tellnes due to different conditions of crystallization.
7.2. Sequence of crystallization: constraints from
experimental studies and phase diagrams
Experimental phase equilibrium data from 1 bar to
13 kb for magmas related to Proterozoic anorthosite
have been presented by Fram and Longhi (1992),
Vander Auwera and Longhi (1994), Vander Auwera et
al. (1998), and Longhi et al. (1999). These phase
diagrams for compositions TJ and HLCA, similar to the
jotunite 12.5V-6S (264) (Table 4), enable us to assess
crystallization conditions for the Tellnes ilmenite
deposit. Indeed, sample 12.5V-6S (264), is interpreted
to represent the trapped liquid component of the ore.
Much evidence also points to this composition being
close to the parental magma of the intrusion. In addition
to compositional similarities with other primitive
jotunites in Rogaland (Table 4), this fine-grained sample
comes from the floor of the ore body, at the contact with
the host anorthosite, exactly where the first injection of
magma would have been quenched.
The progression of liquidus equilibria with pressure for
the Tjörn composition (TJ, Table 4) calculated from the
algorithms of Longhi (1991) is presented in Fig. 11A. The
Opx-Ol liquidus boundary is a peritectic from 1 bar to 5 kb
and becomes a cotectic at 7 kb due to the reduction of the
stability field of olivine with pressure. A similar shift of
the Opx-Ol liquidus boundary is observed in Fig. 11B for
a different magma (HLCA, Table 4) at the same
crystallization pressure of 5 kb. Higher mg-number and
anorthite content of the normative feldspar have a similar
effect on the position of the Opx-Ol liquidus boundary as
increasing pressure (Longhi et al., 1999).
Discrepancies between the evolution in Bjerkreim-
Sokndal (peritectic reaction of orthopyroxene and olivine)
and in Tellnes (cotectic relation between orthopyroxene
and olivine) can be explained by slightly different parental
282 B. Charlier et al. / Chemical Geology 234 (2006) 264–290

Fig. 11. Liquidus equilibria for magmas related to massif-type anorthosites projected from the Pl, Or and Ilm components onto the Ol–Qtz–Wo plane
and calculated from the algorithms of Longhi (1991), modified after Longhi et al. (1999). (A) Progression of liquidus equilibria for TJ-like
compositions (samples 80.12.3A) from 1 bar to 10 kbar. (B) Comparison between liquidus equilibria at 5 kb for HLCA and TJ compositions. (C)
Fractional crystallization path for a TJ parental magma at 5 kb reproducing the crystallization sequence of Bjerkreim-Sokndal. (D) Fractional
crystallization path for a TJ parental magma at 7 kb reproducing the crystallization sequence of Tellnes. (E) Fractional crystallization path for a HLCA
parental magma at 5 kb reproducing the crystallization sequence of Tellnes. (Symbols: pl plagioclase, ol olivine, opx orthopyroxene, cpx
clinopyroxene, ilm ilmenite, gar garnet, sp spinel, sil silica phase).

magma composition and/or by different pressure of em-
placement. At the base of MCU III and IV of Bjerkreim-
Sokndal (Fig. 10), the sequence of crystallization is
plagioclase + olivine (+ilmenite,+magnetite), then pla-
gioclase + orthopyroxene (+ilmenite, −magnetite) fol-
lowed by plagioclase + orthopyroxene + clinopyroxene
(+ilmenite,+magnetite,+apatite). This path is reproduced
in Fig. 11C for fractional crystallization of the TJ parental
magma at 5 kb. The sequence in Tellnes, where plagioclase
(+ ilmenite) is followed by plagioclase+ orthopyroxene +
olivine (+ilmenite) is displayed in Fig. 11D for fractional
crystallization of the TJ parental magma at 7 kb, and in Fig.
11E for fractional crystallization of the HLCA parental
magma at 5 kb. The simultaneous appearance of
orthopyroxene and olivine in Tellnes result from the
position of the melt on the Opx-Ol cotectic. The cotectic
proportion of olivine to orthopyroxene is small which is
consistent with the relatively low abundance of cumulus
olivine in the upper part of Tellnes.
Whether the pressure of crystallization (5 vs. 7 kb;
Fig. 11D) or the composition of the parental magma (Fig.
11E) is responsible for differences in the sequence of
crystallization in the Bjerkreim-Sokndal layered intru-
sion compared to Tellnes may be assessed through the
composition of liquidus phases in these two intrusions
and in experimental products from the crystallization of
TJ and HLCA. The TJ composition (at 5 kb) crystallizes
plagioclase An48 and olivine Fo65 (Vander Auwera and
Longhi, 1994). In Tellnes, the composition of liquidus
plagioclase reaches An56 (see below) and olivine in the
 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
UCZ is around Fo80 (Wilmart et al., 1989) and may reach
Fo84 (Charlier, unpublished data). These values are
slightly more primitive than in the Bjerkreim-Sokndal
layered intrusion were the most primitive compositions
reported by Wilson et al. (1996) are An54 and Fo77. These
values for Tellnes and Bjerkreim-Sokndal are closer to
the composition of liquidus phases produced by the
crystallization of the HLCA composition at 5 kb: An64,
Fo76 (Fram and Longhi, 1992). In Tellnes, the higher
Mg# of the melt when olivine appears in the UCZ
probably results from the fractionation of plagioclase
and ilmenite in the LCZ, both MgO-poor.
Consequently, we consider that the Tellnes parental
magma was more primitive (higher Mg# and more
anorthitic) than the TJ composition of Vander Auwera and
Longhi (1994). These authors also concluded that TJ is
too evolved to be parental to Bjerkreim-Sokndal.
However, the Tellnes parental magma has jotunitic
affinities and is probably close to the marginal sample
12.5V-6S (264) (Table 4) and slightly more primitive than
the Bjerkreim-Sokndal parental magma. The more
primitive character of the Tellnes parental magma is
responsible for the cotectic orthopyroxene–olivine liqui-
dus boundary in Tellnes. The pressure of crystallization is
thus inferred to be the same for Tellnes and Bjerkreim-
Sokndal, around 5 kb.
7.3. Compositional variations of plagioclase and
whole-rocks
A primary cryptic compositional variation of plagio-
clase due to fractional crystallization would be expected
from LCZ to UCZ. Similarly, whole-rock compositions
are believed to be influenced by the change in phase
relations (+orthopyroxene and +olivine). Moreover, the
presence of variable proportions of trapped liquid in
Tellnes is shown by petrographical considerations and by
the range of P2O5 contents. The crystallization of trapped
liquid is known to potentially have a major influence on
the compositions of cumulus minerals (“the trapped liquid
shift” of Barnes, 1986) and of the bulk cumulates.
Plagioclase compositions plotted in various diagrams
(Fig. 7) display well defined linear trends. This feature
can be interpreted as the result of two-component
mixing. One component, called the “primitive pole”, has
the approximate composition An48, Or3.0, 1170 ppm Sr
and 150 ppm Ba. This represents plagioclase from the
most primitive samples in the central zones which have
low bulk P2O5 contents and are interpreted to be the
closest preserved approximation to the primary cumulus
plagioclase composition. The second component, called
the “evolved pole”, has the composition of plagioclase
283
in 12.5V-6S (264): An39.3, Or7.1, 964 ppm Sr, and
355 ppm Ba. As mentioned above, sample 12.5V-6S
(264) is interpreted as representing a liquid composition
with jotunitic affinities. The range of plagioclase
composition can thus be interpreted as a result of
mixing between cumulus plagioclase (“primitive pole”)
and plagioclase that crystallized from the trapped liquid
(“evolved pole”). All intermediate compositions are
observed. Samples from the margins of the ore body are
closer to the “evolved pole” whereas samples from the
centre are nearer to the “primitive pole”. This reflects a
higher proportion of trapped liquid at the margins than
in the centre of the body.
Triangular patterns of whole-rocks composition can
be interpreted similarly but three poles are necessary
(Fig. 9). The first pole has the composition of sample
12.5V-6S (264), and represents the trapped liquid
component of the bulk cumulates. The second pole is
represented by a TiO2-rich cumulate, close to that of
LCZ. This pole thus belong to a group of rocks which
have plagioclase and ilmenite as cumulus minerals (pi-
C), with zero trapped liquid. The third pole is MgO-rich
and is close to cumulates of the UCZ where plagioclase,
ilmenite, orthopyroxene and olivine are cumulus
minerals (piho-C) and the trapped liquid fraction is
zero. Linear trends simply result from the alignment of
the three poles (e.g. TiO2–SiO2, Fig. 9). The composi-
tions of plagioclase in LCZ and in UCZ are not
significantly different (Fig. 6) and plot on a single linear
trend (Fig. 7) because the trapped liquid shift overprints
the effects of fractional crystallization. Samples from the
UCZ do not display a simple linear trend in bulk
cumulate compositions between a pure adcumulate and
the jotunite composition. This results from the increas-
ing proportion of orthopyroxene upwards through the
UCZ (see Table 1).
The composition of these cumulate poles and their
relative proportions of cumulus phases are calculated by
subtracting the effect of the trapped liquid from the bulk
cumulate compositions. This can be done by assuming
that a pure adcumulate does not contain any P2O5. The
intercept for 0% P2O5 gives the composition of the pure
adcumulate. This is illustrated on Fig. 12A for MgO. The
composition and the average modal proportions of
cumulus minerals for the two types of cumulates (pi-C
and piho-C) is given in Table 5. For pi-C, the linear trend
between the adcumulate and the trapped liquid composi-
tions (Fig. 12A) is well-defined which means that the
relative proportion of cumulus plagioclase and ilmenite
(p: 44 wt.%; i: 56 wt.%) remains constant. The variability
of plagioclase/ilmenite proportions in bulk cumulates
results from variable contents of trapped liquid, which has
284 B. Charlier et al. / Chemical Geology 234 (2006) 264–290

7.4. Abundance, role and compositional variation of

trapped liquid

The trapped liquid proportion can be quantitatively

estimated with the following method. As apatite is
interstitial and the P2O5 content of the cumulus minerals
(plagioclase, ilmenite, orthopyroxene and olivine) can be
neglected, we can assume that all P2O5 is present in the
trapped liquid. Knowing that sample 12.5V-6S (264),
which is considered to represent the composition of the
trapped liquid, has 0.69% P2O5, it is possible to calculate
the proportion of trapped liquid in each sample. In bulk
cumulates, P2O5 ranges from 0.16 to 0.58%, cor-
responding to a proportion of trapped liquid between 23
and 83%. Graphically, the amount of trapped liquid can
illustrated in the diagram MgO–P2O5 (Fig. 12A). The
higher proportion of trapped liquid in samples from
marginal zones results from a higher rate of cooling near
the margins with the host anorthosite, preventing the
migration of trapped liquid and resulting in a chilled
margin. The proportion of trapped liquid can also be
calculated by comparing plagioclase compositions with
the P2O5 content of the bulk cumulate (Fig. 12B). In this
case, the presence of trapped liquid decreases the An
content of the bulk plagioclase composition. The
unmodified composition of cumulus plagioclase as
obtained for 0% trapped liquid (0% P2O5) varies between
An47 and An56. This range of plagioclase compositions is
close to the data for the most primitive cumulates of the
Fig. 12. (A) MgO vs. P2O5 for bulk cumulates from the Tellnes
Bjerkreim-Sokndal layered intrusion (Fig. 10), but
ilmenite deposit with the calculated proportion of trapped liquid
assuming 0% liquid component for 0% P2O5 and 100% in the jotunite
representing the trapped liquid composition (TL) (Same legend as Fig.
Table 5
10). (B) An content of plagioclase vs. P2O5 of the bulk cumulates of
Composition and mineral proportions of theoretical adcumulate poles
the Tellnes ilmenite deposit with the calculated proportion of trapped
of the Tellnes ilmenite deposit
liquid assuming 0% liquid component for 0% P2O5 and 100% in the
plagioclase from the jotunite (TL). Pole pi-C Pole piho-C (1) Pole piho-C (2)
Major elements composition of bulk adcumulates (wt.%)
SiO2 23.0 25.0 25.0
a higher plagioclase/ilmenite proportion than the cumulus TiO2 25.3 24.5 23.0
assemblage. The trend is more variable for piho-C, Al2O3 12.5 10.0 8.0
particularly for MgO (Fig. 9). This results from the FeOtot 26.0 27.0 28.0
continuous increase in the modal proportion of orthopyr- MnO 0.12 0.13 0.15
MgO 2.5 5.5 9.0
oxene towards the top of the UCZ. In this case, an increase
CaO 4.5 3.5 3.0
in the rate of orthopyroxene accumulation in the course of Na2O 2.5 2.2 2.0
crystallization is suspected. Two different proportions K2O 0.2 0.2 0.2
have therefore been calculated. The first one, piho-C (1) P2O5 0.0 0.0 0.0
(p: 37 wt.%; i: 52 wt.%; h + o: 11 wt.%) represents the Total 96.62 98.03 98.35
proportions of cumulus phases at the appearance of
Proportions of cumulus minerals (wt.%)a
cumulus orthopyroxene and olivine, and the second one, Plag 46 37 30
piho-C (2) (p: 30 wt.%; i: 49 wt.%; h + o: 21 wt.%), are the Ilm 54 52 49
proportions for the most evolved cumulates observed in Opx + Ol 0 11 21
Tellnes. There is a continuous gradation in modal propor- a
Calculated with Al2O3 in plagioclase, TiO2 in ilmenite; Opx + Ol
tions from the first to the second. by difference.
 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
slightly more calcic. This justifies the interpretation from
phase diagrams that the Tellnes parental magma was
slightly more primitive compared to the Bjerkreim-
Sokndal parental magma.
The spatial distribution of the trapped liquid fraction
is shown in Fig. 13 for section 1600. This trapped
liquid fraction decreases continuously from the margin
to the centre of the ore body. At >100 m from the
contact, the trapped liquid fraction seems to stabilize
around 30%. Following the terminology of Irvine
(1982), cumulates from Tellnes are mainly orthocumu-
lates (the trapped liquid fraction is between 25 and
50%). This is surprising in view of the depth of
emplacement (5 kb) and presumably a relatively slow
rate of crystallization.
Calculations of the trapped liquid fraction have been
done under the assumption that the trapped liquid
maintains the same composition. However, fractional
crystallization causes P2O5 enrichment in the residual
liquid until cumulus apatite appears. So, in more
evolved cumulates, and particularly in piho-C, the
P2O5 content of the magma was higher than in the
magma crystallizing pi-C cumulates. The calculated
proportion of trapped liquid is, therefore, increasingly
overestimated upwards. To estimate the maximum error,
we can consider that the most evolved cumulate at the
top of section 1600 (Fig. 13) crystallized when the
fraction of residual liquid was 0.6, which is the
maximum on the arrival of cumulus apatite by analogy
with Bjerkreim-Sokndal (Duchesne and Charlier, 2005).
For incompatible elements, the liquid composition (Cliq)
Fig. 13. P2O5 in bulk cumulates and trapped liquid fraction (P2O5 in
bulk cumulate/P2O5 in the parental magma) as a function of the depth
(m) in drill core 16V-3S. The dotted line represents the trapped liquid
fraction corrected for the P2O5 increase of the trapped liquid due to
fractional crystallization of the parental magma. Each mining data
corresponds to 6 m of the drill core (number of sample = 41).
285
follows the simplified Rayleigh equation Cliq = C0 /
F = 1.15 (for F = 0.6) where C0 is the P2O5 content
(= 0.69%) in the parental magma. The trapped liquid
fraction in cumulates from the upper part of section
1600 (P2O5 = ∼ 0.22%) is thus 19% and not around 32%.
The calculated trapped liquid fractions with this
correction are shown in Fig. 13.
7.5. Cotectic proportions or sorted cumulates?
One of the main questions concerning the formation
of the ore is how ilmenite was concentrated and why the
proportion of ilmenite to plagioclase is so high. Mineral
sorting is a process that may be responsible for
compositional variations. Therefore it is important to
determine if the relative proportions of cumulus
minerals represent cotectic proportions or not, i.e.
whether the cumulates have been sorted. In Tellnes,
and particularly in the LCZ, variations in whole-rock
compositions (Fig. 9) cannot be explained by a variable
content of one of the cumulus minerals because trends
do not always point towards either plagioclase or
ilmenite. Moreover, whole-rock compositions are al-
ways restricted to well-defined triangles in different
diagrams, and the most important cause of composi-
tional variability seems to be the proportion of trapped
liquid. However, as previously mentioned, the propor-
tion of ilmenite in Tellnes is higher compared to its
proportion in Bjerkreim-Sokndal. Moreover, even if the
experimentally estimated cotectic proportions of Toplis
and Carroll (1995) vary with temperature and oxygen
fugacity, the ilmenite abundance is never higher than
6.5 vol.%. This points to significant ilmenite accumu-
lation in Tellnes where the ilmenite proportion can reach
∼ 40 vol.% in the LCZ. This issue is further discussed
by Charlier et al. (submitted for publication) who have
calculated that the cotectic proportion of ilmenite in
Tellnes is close to 17 wt.% by modelling ilmenite
composition with the Rayleigh equation.
The ore body represents the lower part of a larger
intrusion. This must be borne in mind when considering
possible magma chamber processes. The unexposed
upper part of the ore body may contain a large amount of
plagioclase which could have accumulated due to its
buoyancy in the relatively dense melt (∼2.75, Vander
Auwera et al., in press). However, because the cumulus
plagioclase/ilmenite proportions remain globally con-
stant, the amount of plagioclase removed by flotation
must have been kept constant throughout the course of
crystallization of the exposed part of Tellnes. In other
words, the rate of ilmenite accumulation was consis-
tently higher than the rate of plagioclase accumulation.
286 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
7.6. Emplacement mechanism of the Tellnes ilmenite
deposit
Wilmart et al. (1989) proposed that the ilmenite-rich
norite was injected as a crystal-mush lubricated by 3 to
10% trapped liquid. This mechanism of emplacement was
later supported by magnetic and petrofabric evidence (Diot
et al., 2003). Our new data reveal more complex relation-
ships in major element variation diagrams than the simple
mixing line observed by Wilmart et al. (1989), and exclude
an emplacement mechanism involving the injection of a
crystal-mush. We have shown that the leucocratic pole in
variation diagrams is not an anorthositic cumulate as pro-
posed by Wilmart et al. (1989), but represents the trapped
liquid component. The stratigraphic succession of two
types of cumulate, pi-C followed by piho-C, can hardly be
explained by the injection of a mush but favours in situ
crystallization within the Tellnes magma chamber.
Several points can be discussed concerning the
orientation of the ore based on magnetic and petrofabric
evidence (Diot et al., 2003). These authors concluded that
Tellnes was injected into a zone of weakness in the Åna-
Sira anorthosite, under brittle conditions. However, they
describe irregular, sinuous magmatic contacts on the
northern flank, concordant with the host anorthosite. This
favours an emplacement into ductile anorthosite, corrob-
orated by the irregular and sinuous network of small
dykes in the south-eastern part. Considering the micro-
textures and the criteria proposed by Nicolas (1987) and
Paterson et al. (1989) to distinguish magmatic from solid-
state deformation, plagioclase and orthopyroxene show
evidence of plastic strain in crystals resulting from high-T
solid-state deformation (bending, undulatory extinction,
kinking and partial recrystallization). These deformation
textures could not have been contemporaneous with the
emplacement because there is no relation between the
degree of deformation and recrystallization of a sample
and its trapped liquid fraction (TLF). Samples with >40%
of TLF, much above the critical melt fraction when no
deformation can affect solids (Van der Molen and
Paterson, 1979), can display a higher deformation than
samples with lower TLF. These new considerations imply
that the mineral lineation must be explained by other
mechanisms than by flow of a crystal mush.
We propose that the ilmenite-rich norite crystallized in
situ in a sill-shaped magma chamber with a floor dipping
to the south-east, injected into the Åna-Sira anorthosite
under ductile conditions. Solid-state deformation of the
sill was subsequently produced by subsidence. The direc-
tion of the Tellnes subsidence is relative to the host
anorthosite because the movement can either have been
produced by gravity-induced sinking of the high-density
Tellnes intrusion into the low-density Åna-Sira anortho-
site or/and by the rising of the anorthosite diapir. Actually,
this last movement is probably responsible for the rotation
of the ore body to the NE because the centre of the diapir
was situated SW to the ore body. This explains the vertical
SW flank and why the Lower Central Zone is strati-
graphically higher on the SW flank compared to the NE
flank (Fig. 6).
This subsidence is responsible for deformation of the
minerals and the present shape of the ore body,
producing the magnetic lineation and the 18° SE plunge
of the floor. This is similar to the mechanism described
by Paludan et al. (1994) and Bolle et al. (2002) to explain
deformation of the Bjerkreim-Sokndal layered intrusion.
Our model can be summarized in three successive stages.
First, a jotunitic magma was emplaced in a sill-shaped
magma chamber. As the parental magma cooled, pi-C
accumulated at the bottom of the chamber, followed by
piho-C, with a higher rate of ilmenite accumulation
compared to plagioclase, probably buoyant in the dense
jotunitic melt. The high thermal gradient at the margins
induced formation of rapidly-cooled margins (LMZ and
UMZ). Even in the centre of the ore body, large amounts
of liquid (20–30%) became trapped in the cumulates.
The exposed portion of the deposit represents the lower
part of a larger magma chamber. At the present erosion
level, no evolved cumulates or other rocks that crystal-
lised from residual liquid are preserved. Finally, the
cumulates were deformed by gravity-induced subsi-
dence and by anorthosite diapirism, producing the
present asymmetric trough-shape cross section of the
ore body. This deformation is responsible for the solid-
state recrystallization of minerals and for the present
magnetic fabric orientation.
7.7. The origin of Fe–Ti ores: a synthesis
Several mechanisms have been suggested for the
genesis of ilmenite deposits associated with massif-type
anorthosites. Philpotts (1967) and Kolker (1982) argued
for immiscible Fe–Ti–P-rich melts. Unrealistically high
temperatures for the ilmenite–apatite melts (∼1400 °C)
and experimental studies by Lindsley (2003) do not
favour immiscibility. Furthermore, constant proportions
of mafic minerals in cumulates from the Bjerkreim-
Sokndal layered intrusion (Duchesne and Charlier, 2005)
preclude the formation of an immiscible Fe–Ti rich liquid
during the course of crystallization. The interstitial habit
of ilmenite and the existence of ilmenite-rich veins in
anorthosite are the result of subsolidus grain boundary
modifications and solid-state deformation rather than
liquid immiscibility (Duchesne, 1996).
 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
Robinson et al. (2003) suggested that magma
mixing was responsible for the high proportion of
ilmenite in Tellnes. However, as illustrated throughout
this work, ilmenite is never the only liquidus mineral
but crystallized either together with plagioclase or with
plagioclase, orthopyroxene and olivine. Moreover, the
hypothetical curved ilmenite–plagioclase cotectic in-
voked by these authors, similar to the curved chromite–
olivine cotectic (Irvine, 1977), has no experimental
support. Magma mixing in chambers emplaced in the
continental crust is typically accompanied by system-
atic variations in mineral compositions and isotopic
ratios. The very restricted range of variation of Sr
isotopes in the Tellnes deposit (Table 2) does not
favour such mixing as a mechanism for ilmenite
enrichment.
Finally, the Ti budget is not problematic at all. The
parental magma of Tellnes (sample 12.5V-6S, 264) and
other related melts in the Rogaland Anorthosite Province
have more than 4% TiO2 (Table 4). This is not a feature
that is restricted to Rogaland because TiO2-rich liquids
have been described in other anorthositic complexes (e.g.
Mitchell et al., 1996), and even in modern lavas of
continental tholeiitic suites such as Craters of the Moon or
Snake River Plain (Nekvasil et al., 2000).
It emerges that the only plausible mechanism for the
formation of ilmenite deposits is simply fractional
crystallisation. Parental magmas are Ti-rich and crys-
tallize ilmenite as an early liquidus mineral. Parameters
controlling the sequence of crystallization (e.g. ilmenite
before or after orthopyroxene) and the exact cotectic
proportions are presently not constrained by experi-
mental studies. Further studies on the role of magma
composition, pressure and fO2 are needed. Sorting of
ilmenite due to a higher rate of accumulation compared
to plagioclase is easily explained by different densities
(2.7 vs. 4.8 g cm− 3). Under appropriate conditions, we
suggest that this mechanism can result in a pure ilmenite
body, such as the Lac Tio deposit. This would
preferentially happen in a slowly cooling magma
chamber rather than in a small, rapidly-cooled body of
magma such as at Tellnes.
8. Conclusions
The Tellnes ilmenite deposit is not an homogeneous
ore body. Study of three N–S cross-sections reveals
systematic variations in the cumulus assemblage. In the
lower part (LCZ), plagioclase and ilmenite are the only
cumulus minerals, followed by the appearance of
orthopyroxene and olivine in the upper part (UCZ).
Marginal Zones (UMZ and LMZ) are characterized by
287
high plagioclase and apatite contents and lower propor-
tions of ilmenite. Phase diagrams for magmas related to
massif-type anorthosites reproduce the sequence of
crystallization observed in Tellnes for fractional crystal-
lization of a primitive jotunite at 5 kb. The composition
of the parental liquid for the ore body, identified by the
occurrence at the base of the intrusion of a fine-grained
rock at the contact with the host anorthosite, has a
jotunitic composition slightly more primitive (more
anorthitic and higher Mg #) than that of the Bjerkreim-
Sokndal layered intrusion. Exposed cumulates in Tellnes
correspond to the first stage of evolution of a jotunitic
magma. The ore body is thus interpreted as the lower
part of a larger intrusion.
The proportion of trapped liquid is the main cause of
variability of the compositions of plagioclase and bulk
cumulates. Estimation of the trapped liquid proportion
reveals that it varies from about 20 to 80%. This proportion
is greater at the margins where the cooling rate was higher,
and it continuously decreases towards the centre of the ore
body. Constant Sr isotopes ratios preclude significant
contamination and give no evidence of magma mixing.
The proportion of ilmenite, up to 50 wt.% in the LCZ, is
not cotectic. A higher rate of ilmenite accumulation, due to
plagioclase buoyancy in a dense jotunitic melt is invoked.
The orientation of the minerals and the present trough-
shape of the ore body were acquired during subsidence of
the dense ore body into the Åna-Sira anorthosite and by
up-doming of the anorthosite.
Based on general considerations of the genesis of
ilmenite ores in massif-type anorthosites, it seems that
fractional crystallization of a TiO2-rich magma with
ilmenite as an early liquidus minerals is the most plausible
origin.
Acknowledgements
This work was funded by the Belgian Fund for Joint
Research and the Fund for Research in Industry and
Agriculture. Fieldwork was partly supported by the
Fourmarier Foundation. TITANIA A/S is gratefully
acknowledged for providing access to chemical data on
the Tellnes deposit and for supporting the cost of new
analyses. Kari Berge and Ragnar Hagen are thanked for
their continuous interest and encouragement. We also
acknowledge Kåre Kullerud for sharing unpublished
microprobe data as well as Olivier Bolle and Rais
Latypov for their useful remarks. Comments from Judith
Hannah and Richard Wilson have greatly improved an
earlier version of the manuscript. This paper has
benefited from constructive reviews by Sarah-Jane
Barnes and Brian Robins. [RLR]
288 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
References
Ashwal, L., 1993. Anorthosites. Springer, Heidelberg.
Barling, J., Weis, D., Demaiffe, D., 2000. A Sr-, Nd- and Pb-isotopic
investigation of the transition between two megacyclic units of the
Bjerkreim-Sokndal layered intrusion, south Norway. Chemical
Geology 165, 47–65.
Barnes, S.J., 1986. The effect of trapped liquid crystallization on
cumulus mineral compositions in layered intrusions. Contributions
to Mineralogy and Petrology 93, 524–531.
Barnichon, J.-D., Havenith, H., Hoffer, B., Charlier, R., Jongmans, D.,
Duchesne, J.C., 1999. The deformation of the Egersund Ogna
massif, south Norway: finite element modelling of diapirism.
Tectonophysics 303, 109–130.
Bédard, J.H., 2001. Parental magmas of the Nain plutonic suite
anorthosites and mafic cumulates: a trace element modelling
approach. Contributions to Mineralogy and Petrology 141,
747–771.
Bindeman, I.N., Davis, A.M., Drake, M.J., 1998. Ion microprobe study
of plagioclase–basalt partition experiments at natural concentra-
tion levels of trace elements. Geochimica et Cosmochimica Acta
62, 1175–1193.
Bingen, B., Stein, H.J., 2003. Molybdenite Re–Os dating of biotite
dehydration melting in the Rogaland high-temperature granu-
lites, S. Norway. Earth and Planetary Science Letters 208,
181–195.
Bingen, B., van Breemen, O., 1998. U–Pb monazite ages in amphibolite-
to granulite-facies orthogneiss reflect hydrous mineral breakdown
reactions: Sveconorwegian Province of SW Norway. Contributions
to Mineralogy and Petrology 132, 336–353.
Bingen, B., Demaiffe, D., van Breemen, O., 1998. The 616-My-old
Egersund basaltic dike swarm, SW Norway, and the Late
Proterozoic opening of the Iapetus Ocean. The Journal of Geology
106, 565–574.
Bolle, O., 1996. Mélanges magmatiques et tectonique gravitaire dans
l'Apophyse de l’intrusion de Bjerkreim-Sokndal (Rogaland,
Norvège). Pétrologie, géochimie et fabrique magnétique. Ph.D.
dissertation, Université de Liège.
Bolle, O., Trindade, R.I.F., Bouchez, J.L., Duchesne, J.C., 2002.
Imaging downward granitic magma transport in the Rogaland
Igneous Complex, SW Norway. Terra Nova 14, 87–92.
Bolle, O., Demaiffe, D., Duchesne, J.C., 2003. Petrogenesis of
jotunitic and acidic members of an AMC suite (Rogaland
Anorthosite Province, SW Norway): a Sr and Nd isotopic
assessment. Precambrian Research 124, 185–214.
Bologne, G., Duchesne, J.C., 1991. Analyse des roches silicatées par
spectrométrie de fluorescence X: précision et exactitude. Profes-
sional Papers - Geological Survey of Belgium 249, 1–11.
Charlier, B., Skår, Ø., Korneliussen, A., Duchesne, J.C., Vander
Auwera, J., submitted for publication. Major and trace element
composition of ilmenite in the Tellnes Fe–Ti deposit: fractional
crystallization model and postcumulus evolution.
Demaiffe, D., 1977. Les massifs satellites anorthosito-leuconoritiques
d'Hidra et de Garsaknatt: leur signification pétrogénétique.
Annales de la Société Géologique de Belgique 100, 167–174.
Demaiffe, D., Hertogen, J., 1981. Rare earth element geochemistry and
strontium isotopic composition of a massif-type anorthositic–
charnockitic body: the Hidra Massif (Rogaland, SW Norway).
Geochimica et Cosmochimica Acta 45, 1545–1561.
Diot, H., Bolle, O., Lambert, J.-M., Launeau, P., Duchesne, J.C., 2003.
The Tellnes ilmenite deposit (Rogaland, south Norway): magnetic
and petrofabric evidence for emplacement of a Ti-enriched noritic
crystal mush in a fracture zone. Journal of Structural Geology 25,
481–501.
Duchesne, J.C., 1972. Iron–titanium oxide minerals in the Bjerkrem-
Sogndal Massif, south-western Norway. Journal of Petrology 13,
57–81.
Duchesne, J., 1987. The Bjerkreim-Sokndal Massif. In: Maijer,
C., Padget, P. (Eds.), The Geology of the Southernmost
Norway. Norske Geologiske Undersøkelse Special Publication,
pp. 56–59.
Duchesne, J., 1996. Liquid ilmenite or liquids ilmenite: a comment on
the nature of ilmenite vein deposits. In: Demaiffe, D. (Ed.),
Petrology and Geochemistry of Magmatic Suites of Rocks in the
Continental and Oceanic Crusts. A Volume Dedicated to Professor
Jean Michot. Université Libre de Bruxelles: Royal Museum for
Central Africa, Tervuren, pp. 73–82.
Duchesne, J.C., 1999. Fe–Ti deposits in Rogaland anorthosites (south
Norway): geochemical characteristics and problems of interpreta-
tion. Mineralium Deposita 34, 182–198.
Duchesne, J., 2001. The Rogaland Intrusive Massifs, an Excursion
Guide. NGU report 2001. Geological Survey of Norway 29.
Duchesne, J.C., Charlier, B., 2005. Geochemistry of cumulates
from the Bjerkreim-Sokndal layered intrusion (S. Norway). Part
I: constraints from major elements on the mechanism of
cumulate formation and on the liquid line of descent. Lithos 83,
229–254.
Duchesne, J.C., Hertogen, J., 1988. Le magma parental du lopolithe de
Bjerkreim-Sokndal (Norvège méridionale). Comptes Rendus de
l'Académie des Sciences de Paris 306, 45–48.
Duchesne, J.C., Wilmart, E., 1997. Igneous charnockites and related
rocks from the Bjerkreim-Sokndal layered intrusion (southwest
Norway): a jotunite (hypersthene monzodiorite)-derived A-type
granitoid suite. Journal of Petrology 38, 337–369.
Duchesne, J.C., Roelandts, I., Demaiffe, D., Hertogen, J., Gijbels, R.,
De Winter, J., 1974. Rare-earth data on monzonoritic rocks related
to anorthosites and their bearing on the nature of the parental
magma of the anorthositic series. Earth and Planetary Science
Letters 24, 325–335.
Duchesne, J.C., Roelandts, I., Demaiffe, D., Weis, D., 1985. Petrogenesis
of monzonoritic dykes in the Egersund-Ogna anorthosite (Rogaland,
S.W. Norway): trace elements and isotopic (Sr–Pb) constraints.
Contributions to Mineralogy and Petrology 90, 214–225.
Duchesne, J.C., Wilmart, E., Demaiffe, D., Hertogen, J., 1989.
Monzonorites from Rogaland (southwest Norway): a series of
rocks coeval but not comagmatic with massif-type anorthosites.
Precambrian Research 45, 111–128.
Duchesne, J.C., Liégeois, J.P., Vander Auwera, J., Longhi, J., 1999.
The crustal tongue melting model and the origin of massive
anorthosites. Terra Nova 11, 100–105.
Emslie, R.F., Hamilton, M.A., Thériault, R.J., 1994. Petrogenesis of a
Mid-Proterozoic anorthosite–mangerite–charnockite–granite
(AMCG) complex: isotopic and chemical evidence from the
Nain plutonic suite. Journal of Geology 102, 539–558.
Fram, M., Longhi, J., 1992. Phase equilibria of dikes associated with
Proterozoic anorthosite complexes. American Mineralogist 77,
605–616.
Gierth, E., 1970. Die Ilmenitlagerstätte Tellnes (Süd-Norwegen). Ph.D.
dissertation, T.U. Clausthal.
Gierth, E., 1983. Types and trace elements of titanium ores in the
Tellnes area south-Rogaland. Monograph Series on Mineral
Deposits 22, 47–57.
Gierth, E., Krause, H., 1973. Die ilmenitlagerstätte Tellnes (Süd-
Norwegen). Norsk Geologisk Tidsskrift 53, 359–402.
 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
Gierth, E., Krause, H., 1974. Contribution to the mineralogy of Norway,
No. 57: Baddeleyit von Tellnes. Norsk Geologisk Tidsskrift 54,
193–197.
Gierth, E., Krause, H., Schott, W., 1982. Paragenesen und gefüge des
hydrothermalen nachhalls in südwestnorwegischen titaneisener-
zen. Erzmetall 35, 441–446.
Grove, T.L., Baker, M.B., Kinzler, R.J., 1981. Coupled CaAl–NaSi
diffusion in plagioclase feldspar: experiments and application to
cooling rate speedometry. Geochimica et Cosmochimica Acta 58,
2113–2121.
Hunter, R.H., 1996. Texture development in cumulate rocks. In:
Cawthorn, R.G. (Ed.), Layered Intrusions. Elsevier, Amsterdam,
pp. 77–101.
Irvine, T.N., 1977. Origin of chromitite layers in the Muskox intrusion
and other stratiform intrusions: a new interpretation. Geology 5,
273–277.
Irvine, T.N., 1982. Terminology for layered intrusions. Journal of
Petrology 23, 127–162.
Jensen, J.C., Nielsen, F.M., Duchesne, J.C., Demaiffe, D., Wilson, J.
R., 1993. Magma influx and mixing in the Bjerkreim-Sokndal
layered intrusion, south Norway: evidence from the boundary
between two megacyclic units at Storeknuten. Lithos 29,
311–325.
Jensen, K.K., Wilson, J.R., Robins, B., Chiodoni, F., 2003. A sulphide-
bearing orthopyroxenite layer in the Bjerkreim-Sokndal Intrusion,
Norway: implications for processes during magma-chamber
replenishment. Lithos 67, 15–37.
Kolker, A., 1982. Mineralogy and geochemistry of Fe–Ti oxide and
apatite (nelsonite) deposits and evaluation of the liquid immisci-
bility hypothesis. Economic Geology 77, 1146–1148.
Krause, H., Pape, H., 1977. Untersuchungen zum geologischen und
petrographischen aufbau des Storgangen-ilmeniterzkörpers und
seiner nebengesteinseinheiten (Süd-Norwegen). Norsk Geologisk
Tidsskrift 57, 263–284.
Krause, H., Zeino-Mahmalat, R., 1970. Untersuchungen an erz und
nebengestein der grube Blåfjell in SW-Norwegen. Norsk Geolo-
gisk Tidsskrift 50, 45–88.
Krause, H., Gierth, E., Schott, W., 1985. Fe–Ti deposits in the south
Rogaland igneous complex, especially in the anorthosite-massif of
Åna-Sira. Norges Geologiske Undersøkelese 402, 25–37.
Kress, V.C., Carmichael, I.S.E., 1991. The compressibility of silicate
liquids containing Fe2O3 and the effect of composition, temperature,
oxygen fugacity and pressure on their redox states. Contributions to
Mineralogy and Petrology 108, 82–92.
Kullerud, K., 2003. Geochemistry and mineralogy of the Tellnes ore
body. Norges Geologiske Undersøkelse Special Publication 9, 70–71.
Lafrance, B., John, B.E., Scoates, J.S., 1996. Syn-emplacement
recrystallization and deformation microstructures in the Poe
Mountain anorthosite, Wyoming. Contributions to Mineralogy
and Petrology 122, 431–440.
Lindsley, D.H., 2003. Do Fe–Ti oxide magmas exist? Geology: yes;
experiments: no! Norges Geologiske Undersøkelse Special
Publication 9, 34–35.
Longhi, J., 1991. Comparative liquidus equilibria of hypersthene-normative
basalts at low pressure. American Mineralogist 76, 785–800.
Longhi, J., 2005. A mantle or mafic crustal source for Proterozoic
anorthosites? Lithos 83, 183–198.
Longhi, J., Vander Auwera, J., Fram, M., Monthieth, J.N., 1993.
Pressure effects, kinetics and rheology of anorthositic and related
magmas. American Mineralogist 78, 1016–1030.
Longhi, J., Vander Auwera, J., Fram, M.S., Duchesne, J.C., 1999.
Some phase equilibrium constraints on the origin of Proterozoic
289
(massif) anorthosites and related rocks. Journal of Petrology 40,
339–362.
Maijer, C., 1987. The metamorphic envelope of the Rogaland intrusive
complex. In: Maijer, C., Padget, P. (Eds.), The Geology of the
Southernmost Norway. Norske Geologiske Undersøkelse Special
Publication, pp. 68–72.
Michot, P., 1965. Le magma plagioclasique. Geologische Rundschau
54, 956–976.
Mitchell, J.N., Scoates, J.S., Frost, C.D., Kolker, A., 1996. The
geochemical evolution of anorthosite residual magmas in the
Laramie Anorthosite Complex, Wyoming. Journal of Petrology 37,
637–660.
Möller, A., O'Brien, P.J., Kennedy, A., Kröner, A., 2002. Polyphase
zircon in ultrahigh-temperature granulite (Rogaland, SW Norway):
constraints for Pb diffusion in zircon. Journal of Metamorphic
Geology 20, 727–740.
Nekvasil, H., Simon, A., Lindsley, D.H., 2000. Crystal fractionation
and the evolution of intra-plate hy-normative igneous suites:
insight from their feldspars. Journal of Petrology 41, 1743–1757.
Nicolas, A., 1987. Principles of Rock Deformation. D. Reidel,
Dordrecht.
Nielsen, F.M., Wilson, J.R., 1991. Crystallization processes in the
Bjerkreim-Sokndal layered intrusion, south Norway: evidence
from the boundary between two macrocyclic units. Contributions
to Mineralogy and Petrology 107, 403–414.
Nielsen, F.M., Campbell, I.H., McCulloch, M., Wilson, J.R., 1996.
A strontium isotopic investigation of the Bjerkreim-Sokndal
layered intrusion, southwest Norway. Journal of Petrology 37,
171–193.
Paludan, J., Hansen, U.B., Olesen, N.Ø., 1994. Structural evolution of
the Precambrian Bjerkreim-Sokndal intrusion, south Norway.
Norsk Geologisk Tidsskrift 74, 185–198.
Paterson, S.R., Vernon, R.H., Tobisch, O.T., 1989. A review of criteria
for the identification of magmatic and tectonic foliation in
granitoids. Journal of Structural Geology 11, 349–363.
Philpotts, A.R., 1967. Origin of certain iron–titanium oxide and apatite
rocks. Economic Geology 62, 303–315.
Phinney, W.C., 1992. Partition coefficients for iron between
plagioclase and basalt as a function of oxygen fugacity:
implications for Archean and lunar anorthosites. Geochimica et
Cosmochimica Acta 56, 1885–1895.
Robins, B., Tumyr, O., Tysseland, M., Garman, L.D., 1997. The
Bjerkreim-Sokndal layered intrusion, Rogaland, SW Norway:
evidence from marginal rocks for a jotunite parent magma. Lithos
39, 121–133.
Robinson, P., Kullerud, K., Tegner, C., Robins, B., McEnroe, S.A.,
2003. Could the Tellnes ilmenite deposit have been produced by
in-situ magma mixing? Norges Geologiske Undersøkelse Special
Publication 9, 107–108.
Schärer, U., Wilmart, E., Duchesne, J.C., 1996. The short duration and
anorogenic character of anorthosite magmatism: U–Pb dating of
the Rogaland complex, Norway. Earth and Planetary Science
Letters 139, 335–350.
Scoates, J.S., Chamberlain, K.R., 1997. Orogenic to post-orogenic
origin for the 1.76 Ga Horse Creek anorthosite complex,
Wyoming, USA. Journal of Geology 105, 331–343.
Tegner, C., 1997. Iron in plagioclase as a monitor of the differentiation
of the Skaergaard intrusion. Contributions to Mineralogy and
Petrology 128, 45–51.
Tegner, C., Wilson, J.R., Robins, B., 2005. Crustal assimilation in
basalt and jotunite: constraints from layered intrusions. Lithos 83,
299–316.
290 B. Charlier et al. / Chemical Geology 234 (2006) 264–290
Toplis, M.J., Carroll, M.R., 1995. An experimental study of the
influence of oxygen fugacity on Fe–Ti oxide stability, phase
relations, and mineral-melt equilibria in ferro-basaltic systems.
Journal of Petrology 36, 1137–1170.
Vander Auwera, J., Longhi, J., 1994. Experimental study of a jotunite
(hypersthene monzodiorite): constraints on the parent magma
composition and crystallization conditions (P, T, fO2) of the
Bjerkreim-Sokndal layered intrusion (Norway). Contributions to
Mineralogy and Petrology 118, 60–78.
Vander Auwera, J., Longhi, J., Duchesne, J.C., 1998. A liquid line of
descent of the jotunite (hypersthene monzodiorite) suite. Journal of
Petrology 39, 439–468.
Vander Auwera, J., Longhi, J., Duchesne, J.C., 2000. The effect of
pressure on DSr (plag/melt) and DCr (opx/melt): implications for
anorthosite petrogenesis. Earth and Planetary Science Letters 178,
303–314.
Vander Auwera, J., Weis, D., Duchesne, J.C., in press. Marginal mafic
intrusions as indicators of downslope draining of dense residual
liquids in anorthositic diapirs. Lithos.
van der Molen, I., Paterson, M.S., 1979. Experimental deformation of
partially-melted granite. Contributions to Mineralogy and Petrol-
ogy 70, 299–318.
Wilmart, E., Demaiffe, D., Duchesne, J.C., 1989. Geochemical
constraints on the genesis of the Tellnes ilmenite deposit,
southwest Norway. Economic Geology 84, 1047–1056.
Wilson, J.R., Overgaard, G., 2005. Relationship between the Layered
Series and the overlying evolved rocks in the Bjerkreim-Sokndal
intrusion, southern Norway. Lithos 83, 277–298.
Wilson, J.R., Robins, B., Nielsen, F.M., Duchesne, J.C., Vander
Auwera, J., 1996. The Bjerkreim-Sokndal layered intrusion,
southwest Norway. In: Cawthorn, R.G. (Ed.), Layered Intrusions.
Elsevier, Amsterdam, pp. 231–255.